TIBTEC 1879 No.

of Pages 22

Trends in Biotechnology

Feature Review

Enabling Technologies for Personalized and

Precision Medicine
Dean Ho,1,2,3,4,* Stephen R. Quake,5,6,7 Edward R.B. McCabe,8,23,* Wee Joo Chng,9,10 Edward K. Chow,4,10
Xianting Ding,11 Bruce D. Gelb,12 Geoffrey S. Ginsburg,13 Jason Hassenstab,14,15 Chih-Ming Ho,16
William C. Mobley,17 Garry P. Nolan,18 Steven T. Rosen,19 Patrick Tan,20 Yun Yen,21 and Ali Zarrinpar22

Individualizing patient treatment is a core objective of the medical field. Reaching Highlights
this objective has been elusive owing to the complex set of factors contributing Engineering approaches to precision
to both disease and health; many factors, from genes to proteins, remain un- medicine will harness population-wide
data to identify individualized treatment
known in their role in human physiology. Accurately diagnosing, monitoring, strategies.
and treating disorders requires advances in biomarker discovery, the subse-
quent development of accurate signatures that correspond with dynamic Personalized medicine harnesses a
disease states, as well as therapeutic interventions that can be continuously op- subject’s own data to individualize their
own care, from diagnosis through treat-
timized and modulated for dose and drug selection. This work highlights key ment selection and monitoring.
breakthroughs in the development of enabling technologies that further the
goal of personalized and precision medicine, and remaining challenges that, Novel clinical trial designs will play a vital
when addressed, may forge unprecedented capabilities in realizing truly individ- role in assessing the efficacy and safety
of emerging therapies and diagnostics.
ualized patient care.
Artificial intelligent platforms will globally
Converging Genomic and Phenotypic Medicine with Technology optimize combination therapy from the
Personalized and precision medicine seek to build a foundation for actionable health manage- preclinical through clinical stages of
validation.
ment through a broad spectrum of information. Potential inputs for advancing precision medicine
include longitudinal tracking of healthy individuals to better understand the transition from The widespread deployment of precision
nondiseased to diseased states; more precisely identifying individuals at risk for disease; and tai- and personalized medicine technologies
will involve the convergence of several
loring treatments based on diverse and growing data sets from both individual trials and
factors ranging from evolving education
population-based studies [1]. The data flowing into precision medicine will come from genetic da- at the interface of engineering and
tabases, medical records, tissue banks, and other clinical sources of ‘big data’. In parallel with our medicine and policies that support new
ability to gather an impressive amount of data from any given patient is the expansion of comput- clinical trial designs, to scaling the use
of electronic medical records (EMR) to
ing power and, with it, our analytical capabilities and ability to link data sets together to ‘make
drive clinical decision support.
sense’ of big data. Precision medicine relies on large quantities of population-level data to help
determine the appropriate treatment for an individual. Precision medicine, in some respects,
deviates from traditional medicine by providing insight into how population-derived genetic,
proteomic, or broader biomarker profiles can collectively determine the course of treatment for
1
an individual (Box 1). We should think of human disease as involving ‘networks’ of aberrant The N.1 Institute for Health (N.1),
National University of Singapore,
activities, such as signaling and cell behavior, as well as modifier genes, ‘moonlighting’ proteins, Singapore
etc., rather than as a linear progression [2]. In this way, we might uncover new linkages between 2
The Institute for Digital Medicine
seemingly disparate diseases that open doors to novel treatment approaches. Precision medi- (WisDM), National University of
Singapore, Singapore
cine is already using different sources of patient data to define these new linkages, whether it 3
Department of Biomedical Engineering,
be through genomics in type 2 diabetes or imaging phenotypes in glioblastoma, for instance NUS Engineering, National University of
[2–4]. There are notable challenges to precision medicine that include data accrual, access to pa- Singapore, Singapore
4
Department of Pharmacology, Yong
tients, and data analysis, which we will address in subsequent sections. Loo Lin School of Medicine, National
University of Singapore, Singapore
5
Personalized medicine differs from precision medicine. In personalized medicine, which has been Department of Bioengineering, Stanford
University, CA, USA
considered for some time, the patient is viewed as an individual from diagnosis to therapy, which 6
Department of Applied Physics,
clearly has some overlaps with precision medicine in execution, but unlike precision medicine Stanford University, CA, USA

Trends in Biotechnology, Month 2019, Vol. xx, No. xx https://doi.org/10.1016/j.tibtech.2019.12.021 1

© 2019 Elsevier Ltd. All rights reserved.
Trends in Biotechnology

7
Box 1. Defining Precision and Personalized Medicine Chan Zuckerberg Biohub,
San Francisco, CA, USA
From an engineering perspective, precision medicine involves the use of technologies to acquire and validate population- 8
March of Dimes Foundation,
wise data, such as omics-based single-cell analysis and biomarker discovery, for subsequent application on the individual
New York, USA
patient level. 9
Department of Haematology and
Oncology, National University Cancer
Personalized medicine involves the use of technologies to seriously acquire and assess an individual’s own data for only Institute, National University Health
their own treatment. For example, this may involve the use of artificial intelligence (AI) to both design a drug combination System, Singapore
based on a patient’s own biopsy and follow with N-of-1 dosing protocols. 10
Cancer Science Institute of Singapore,
National University of Singapore,
Broad deployment of both approaches would rely on their successful integration, with genome-guided drug pairing (driven Singapore
by population data) followed by AI-guided dynamic dosing (driven by individual data). 11
Institute for Personalized Medicine,
Shanghai Jiao Tong University,
Shanghai, China
12
Mindich Child Health and Development
does not inherently rely on large data sets or population-based approaches to redefine disease Institute, Departments of Pediatrics and
(Box 1). In personalized medicine, engineers have found a stronghold. Engineers have developed Genetics and Genomic Sciences, Icahn
School of Medicine at Mount Sinai,
enabling technologies ranging from micro/nanofluidics for single circulating tumor cell (CTC) NY, USA
13
(see Glossary) analysis to nanotechnology for isolating extracellular vesicles and exosomes Center for Applied Genomics and
Precision Medicine, Duke University,
from liquid biopsies to imaging platforms to predict the effectiveness of nanomedicines [5,6].
NC, USA
These technologies have the opportunity to provide unprecedented diagnostic insight on a per- 14
Department of Neurology, Washington
sonalized level, but are still making their way into routine clinical care. In particular, personalized University in St. Louis, MO, USA
15
Psychological & Brain Sciences,
pharmacokinetics has enabled new intratumoral devices and microfluidic constructs available
Washington University in St. Louis,
to track an individual’s response to a particular drug [7,8]. The recent introduction of phenotypic MO, USA
16
personalized medicine (PPM) [the harnessing of augmented artificial intelligence (AI) to Department of Mechanical
personalize combination therapy and improve efficacy and safety on the basis of measured Engineering, University of California,
Los Angeles, CA, USA
end-point phenotypes for specific patients] has enabled continuous, patient-specific optimization 17
Department of Neurosciences,
of monotherapy and combination therapy as well as the agnostic design of novel, optimized, University of California, San Diego,
fixed-dose drug combinations [9,10]. Pilot clinical trials involving PPM have been launched for tu- CA, USA
18
Department of Microbiology &
berculosis, HIV (NCT02632474), liver and kidney transplant immunosuppression, hematologic Immunology, Stanford University,
cancers, and other indications, demonstrating the power of engineering platforms to cut CA, USA
19
across medicine. To enable patient-specific treatment, a growing pool of technologies is enabling Comprehensive Cancer Center and
Beckman Research Institute, City of
individualized monitoring. These include everything from wearables, like glucose-monitoring, Hope, CA, USA
20
neurostimulation, neuromonitoring, and sweat-monitoring electronic tattoos (representing Duke-NUS Medical School, National
‘small data’ personalized diagnostics and analytics) to AI-based drug selection and administra- University of Singapore, Singapore
21
College of Medical Technology, Center
tion; to patient-tailored gene-editing nanoparticles that can both image (diagnostic) and treat of Cancer Translational Research, Taipei
(therapy), sometimes called ‘theranostics’ [11–15]. As these data sources grow in scale as well Cancer Center of Taipei Medical
University, Taipei, Taiwan
as quality, being richer and more diverse, medicine will be able to more effectively individualize 22
Department of Surgery, Division of
therapy on the basis of population-wide genetic and phenotypic information, with the hope of Transplantation & Hepatobiliary Surgery,
moving beyond medicine’s ‘one size fits all’ modus operandi. University of Florida, FL, USA
23
Retired

Bioengineering underpins the implementation of precision and personalized medicine and related en-
abling technologies in the clinic (Table 1), as knowledge obtained from traditional biomedical fields
*Correspondence:
makes its way into clinical solutions via engineering tools and approaches. Engineers will play an es- [email protected] (D. Ho) and
pecially crucial role in advancing both personalized and precision medicine, although they may not yet [email protected]
realize how their enabling technologies offer solutions to many unmet clinical needs. In this paper, we (E.R.B. McCabe).

explore recent advances in enabling technologies and state-of-the-art bioengineering approaches

that will be instrumental in not only continuing to advance personalized and precision medicine into
the clinic, but also catalyze its transformation into the standard of care. We also discuss policy con-
siderations that will impact the development, implementation, and adoption of such technologies.

Together, personalized and precision medicine, enabling technologies, and smart policy choices
will open the doors to optimized targeted therapies, to disease prevention, and to an overall shift
in how we think about human disease.

2 Trends in Biotechnology, Month 2019, Vol. xx, No. xx

Trends in Biotechnology

Enabling Technologies for Precision Medicine Glossary

The field of precision medicine has already provided important insights into the mechanisms at Artificial intelligence (AI): in the
play at disease onset; into biological targets that can directly inhibit disease progression; and context of healthcare, AI uses algorithms
into biomarkers that reflect treatment response. Such understanding has collectively mediated to reconcile complex data in an effort to
identify actionable strategies for many
substantial advances towards improving patient treatment outcomes [16–18]. With new data applications. These range from
sources and combinations of these data, precision medicine will propel existing drug-selection improving treatment outcomes to
platforms like pharmacogenomics and patient-derived primary cultures [19–25]. When coupled accelerating drug discovery, among
others.
with big data platforms, these approaches can identify targeted therapies that may predict
Bayesian decision analysis (BDA):
and/or induce improved response rates over clinical standards. Importantly, these advances with regards to healthcare, BDA is used
are being actively assessed in a clinical setting (NCT02795156, NCT03903835). to correlate tradeoffs and decision-
making processes. For example, using
BDA towards novel clinical trial designs
Several clinical trials are currently in place to bring precision medicine to the bedside. One study of
may involve the correlation of outcome
lung cancer patients is pairing genomic analysis of murine and human specimens and coupling objectives for a patient with the benefits
these data with imaging analysis in a co-clinical trial to identify genomic signatures to improve and risks undergoing treatment.
liquid biopsies (NCT02597738). A precision-guided study for treating cancer pain is being con- Chimeric antigen receptor T (CAR-T)
cell immunotherapy: this form of
ducted, where patients are screened for the cytochrome p450 2D6 (CYP2D6) and μ-opioid re-
immunotherapy modifies a patient's
ceptor (OPRM1) genotypes to monitor their response to opioid treatment (NCT02664350). own T cells, which are derived from their
Precision medicine is being used to identify genomic and molecular markers to predict patient immune system, with chimeric antigen
response to the administration of 17 hydroxyprogesterone caproate (17OHPC) as a potential receptors (CAR) on their surfaces. These
modified T cells can then selectively
intervention to prevent preterm birth (NCT02173210). Lastly, one study is combining four predic- target surface markers on the cancer
tive markers for breast cancer (HER2; TP53, CHEK2, and RB1) with massive parallel genetic cells using these receptors during
screening to develop datasets that can be leveraged to individualize treatment regimens treatment.
(NCT02624973). Circulating tumor cell (CTC): this cell
is released by a primary tumor into the
circulatory system and may serve as a
Precision medicine, however, goes beyond genomic medicine. Instead, it is a ‘confluence of bio- foundation for metastasis.
logical, physical, engineering, computer, and health sciences…toward data-driven, mechanism- Clinical decision support (CDS):
using a broad spectrum of applicable
based health and health care for each individual’ [26]. As such, precision medicine looks at a host
data, CDS platforms provide actionable
of data across a population and defines, on the basis of areas like big data, patient response to a guidance to clinicians in areas such as
specific disease state and resulting interventional strategies. It is here that engineering plays a drug selection, dosage modifications,
powerful role in linking these different ‘layers’, to redefine disease and search for meaningful and other courses of treatment.
Clustered regularly interspaced
mechanistic underpinnings that can inform next-generation therapies [26]. We see engineers
short palindromic repeats
continuing to contribute to precision medicine the in the following ways: by enabling biomarker (CRISPR): the CRISPR and CRISPR-
discovery, by creating diagnostic and sensing platforms, and by devising novel drug delivery associated protein 9 (CRISPR-Cas9)
methods to get precision therapeutics to the patient (Figure 1) [27–37]. platform is used for genome editing,
where genetic material can be added,
removed, or modified. This approach
Biomarker Discovery can potentially be used to address a
Engineering approaches to biomarker discovery and validation have resulted in the development multitude of diseases by altering the
of signatures that may eventually serve as dynamic indicators of patient response to therapeutic genetic information that drives the onset
of these disorders.
intervention [38–43]. The genomics revolution and advent of approaches such as next-generation
CURATE.AI: this mechanism-
sequencing (NGS) has played a major role towards broad implementation of precision medicine in independent artificial intelligence
the clinic [44,45]. The fruition of this capability would be vital to enabling individualized treatment platform is used to dynamically optimize
regimens in a patient-specific manner. For example, single molecule array (SiMoA) technology, clinical combination therapy dosing
during the course of treatment. By using
which allows for single molecules to be sequestered by paramagnetic beads for digital readouts
only a patient’s own data to manage
in microwells, was able to detect changes in prostate-specific antigen (PSA) levels that spanned their own combination therapy regimen,
orders of magnitude between individual prostate cancer cells [46]. Fiber microarrays have also CURATE.AI can maximize treatment
been produced for multiplexed diagnostics. In these microarrays, a solution of microbeads that efficacy and safety for a sustained
duration on an individualized basis. It is
contains analogs for different target molecules can be interrogated with an imaging fiber, with
broadly applicable towards oncology,
etched microwells in the fiber providing positive signals for successfully detected biomarkers infectious disease, and many other
[47]. This platform was used to develop a potential signature for cystic fibrosis (CF) using saliva. disease indications.
Via the fiber microarrays, CF patients were observed to have substantially elevated vascular

Trends in Biotechnology, Month 2019, Vol. xx, No. xx 3

Trends in Biotechnology

endothelial growth factor (VEGF), interferon-gamma inducible protein (IP-10), interleukin-8 (IL-8), Cytometry by time of flight (CyTOF):
this is a mass spectrometry
and epidermal growth factor (EGF) levels, while matrix metallopeptidase-9 (MMP-9) levels were methodology that uses heavy metal
lower. The six-biomarker signature, captured by the highly sensitive and multiplexed fiber micro- antibody tags for cell surface and
array, was capable of differentiating patient subgroups and correlated well with the forced expi- intracellular markers. CyTOF analysis
ratory volume (FEV1) readout for disease severity [47]. In addition, long-term monitoring of enables multiplexed profiling of single-
cell responses for applications in drug
inflammatory cytokines in healthy individuals was conducted, revealing a subset of cytokines development and fundamental studies
that can vary by as much as two orders of magnitude between subjects, while others had sub- into cellular mechanisms.
stantially lower levels of interpatient variability [48]. This study demonstrated the importance of Electronic medical records (EMR):
taking baseline studies of patient biomarker signatures as a way towards highly accurate monitor- electronic medical records can contain a
broad spectrum of information
ing of the dynamic states of disease progression and/or treatment response. pertaining to a patient’s health history.
They can serve as vital platforms for the
Microfluidic technologies have a long history of providing the biomedical community unique, implementation of treatment and
diagnostic paradigms that may integrate
modular platforms for detecting low levels of biomarker in small volumes of fluid. In one instance,
emerging technologies such as artificial
a microfluidic device was used to deplete hematopoietic stem cells (HSCs), thereby allowing intelligence, wearables, and other
CTCs associated with hepatocellular carcinoma (HCC) to be easily detected. With a pure sample, modalities.
an RNA-based signature was developed using 10 liver-based transcripts, subsequently resulting Machine learning (ML): ML platforms
use algorithms that are trained with a set
in the detection of HCC-derived CTCs in 9 of 16 untreated patients with HCC as well as 1 of 31
of data to subsequently make inferences
patients that had nonmalignant liver disease. It is important to note that the digital scoring process and identify a course of action without
also did not correlate to commonly monitored serum alphafetoprotein (AFP) levels, demonstrating requiring a directed set of instructions.
the importance of the RNA-based signature in accurately detecting and assessing disease states Implementation of ML typically requires
minimal human interaction. In the
[49]. This RNA signature would not have been possible without the sorting and purifying capabil- context of healthcare, it can be used for
ities of the microfluidic device. many applications, including the design
of drug combinations and the
In addition to the aforementioned approaches, a number of studies have harnessed single- development of biomaterials, among
others.
cell analysis using approaches such as Raman microspectroscopy and multiplexed imaging Maximum tolerated dose (MTD): the
using laser particles for biomarker discovery [50–52]. Peptide arrays are now capable of highest dose of a drug that can be
monitoring low-affinity protein–ligand interactions that may not have been captured by administered to a subject while
simultaneously avoiding an
conventional methodologies, such as Fourier-transform infrared spectroscopy or surface
unacceptable level of toxicity. With
plasmon resonance spectroscopy, opening doors to novel monitoring and drug discovery regards to precision and personalized
strategies [53,54]. medicine, emerging studies have shown
promise in identifying lower drug doses
Precision Diagnostics and Biosensing that result in improved efficacy and
safety, potentially avoiding the need to
A key barrier that confronts precision medicine is the inability, in many cases, to assess quantifi- reach the MTD during therapy.
able treatment outcomes due to inadequate biomarkers. This is especially true for solid cancers Mitochondrial replacement therapy
and certain infectious diseases, such as tuberculosis, and for other indications, such as pain and (MRT): this approach is used to
brain injury. Recent capabilities pertaining to individualized care could especially benefit from address mitochondrial diseases by
replacing mitochondria that contain DNA
frequent biomarker analysis in order to prescribe accurately calibrated treatment responses. mutations with healthy mitochondria. In
Emerging technologies will open the doors to biomarker surveillance with high sensitivity and the context of reproductive medicine, a
specificity, as well as greater frequency than conventional monitoring approaches such as imag- mother with mitochondrial disease can
have her eggs transferred to a donor egg
ing [computed tomography (CT), magnetic resonance imaging (MRI)] or blood draws. For exam-
with healthy mitochondria.
ple, microfluidics and nanofluidics can process and analyze in parallel nucleic acids, proteins, Nanodiamond (ND): nanodiamonds
metabolites, and single cells in various biological fluids, such as urine, blood, and saliva, which in- are carbon-based nanoparticles that
creases the frequency of outcome monitoring, a valuable resource for personalized dosing and can be used to carry multiple classes of
therapeutic and imaging compounds.
pharmacokinetics. In fact, guided therapy using microfluidics or nanofluidics for genomic analysis
Their unique surface electrostatic
and CTC isolation and characterization for biomarker discovery is starting to make its way into the properties have been used to markedly
clinic. CTCs have been isolated using microfluidics from lung adenocarcinoma, undifferentiated improve magnetic resonance imaging
Epstein-Barr virus, positive nasopharyngeal carcinoma, breast cancer, colorectal cancer, contrast efficiency as well as drug
delivery efficacy.
prostate cancer, and gastric cancer to develop predictive biomarker panels for the downstream
Phenotypic personalized medicine
selection of therapeutic regimens (NCT01022723). Circulating endothelial cells and CTCs are (PPM): this artificial intelligence-based
also being evaluated as biomarkers for renal cancer (NCT02499458). A study to capture and approach uses quantifiable measures of

4 Trends in Biotechnology, Month 2019, Vol. xx, No. xx

Trends in Biotechnology

analyze CTCs from metastatic breast cancer patients using antibody panels and microfluidics is clinical efficacy and safety, such as
tumor burden through imaging or
also being conducted (NCT02904135) [5,55–61]. circulating biomarker analysis, as well as
toxicity panels to guide drug dosing. This
Surprisingly little is known about the evolution, origin, and metastasis of cancer. Single-cell anal- approach can be implemented in a
ysis, an area that is maturing rapidly, promises to deliver new disease insights by parsing out the mechanism-independent manner.
Quadratic phenotypic optimization
potential role of cellular heterogeneity in health and disease, including cancer [62,63] (Figure 1). platform (QPOP): this AI-based
For example, digital real-time PCR for single-cell transcription factor profiling in HSCs revealed approach simultaneously identifies the
the presence of two HSC subpopulations from a specific type of progenitor cell that were believed right drugs and corresponding doses
to be homogeneous. This finding made it clear that gene expression could be assessed in a copy- from large pools of candidate therapies
for novel drug combination
number-per-cell context, providing extraordinary insight into transcriptional regulation [57]. development. It can be implemented
Single-cell genomics has also been used to analyze the clonal structure of tumors in both colon without disease target/mechanism
cancer [64] and leukemia [56,65] and the literature using this approach is growing by leaps and information and does not rely on drug
synergy predictions to optimize
bounds.
treatment outcomes.
Spherical nucleic acids (SNA): these
Engineers have found ways to peer into biological processes, sometimes in ‘real-time’, to uncover nanostructures consist of precisely
novel insights into human disease progression. For instance, real-time imaging of colon cancer positioned and high-density
configurations of nucleic acids that have
cell migration in mice has revealed clonal liver metastasis, providing valuable insight into early de-
been explored for gene regulation with
tection and subsequent treatment of advanced cancers [66]. Imaging dormant breast cancer broad applications across different
cells in mice has shown that vascular triggers are responsible for proliferation and growth, reveal- disease indications. They are currently
ing new therapeutic targets in the vasculature [67]. Devices can be used to isolate rare cell pop- being evaluated at the clinical level.
Stamped assembly of polymer
ulations, in some instances to understand individual cell contributions to disease, in others, to layers (SEAL): this approach uses 3D
make a diagnosis or therapeutic decision. In a recent study using blood samples obtained from printed microwells that contain multiple
metastatic castration-resistant prostate cancer patients, the presence of androgen-receptor drugs and can be used for the timed
splice variant 7 (AR-V7) on the CTC surface predicted improved response to taxane therapy release of multiple therapies in a
sustained fashion.
over aminoacyl-tRNA synthetase (ARS) inhibition. Of note, these systems used antibody cocktails Zinc finger nuclease (ZFN): this
to label and distinguish CTCs. Digital pathology algorithms were then used to track cellular mor- enzyme is comprised of DNA-binding
phology from highly heterogeneous solutions in order to classify CTC subtypes. Pathway activa- and cleavage domains and is used as a
genome editing platform. ZFN-based
tion could also be assessed through the additional detection of surface markers in multiple
genome editing therapies are currently
myeloma [68,69]. In addition, this integrated approach was capable of isolating both single being evaluated at the clinical level.
CTCs and clusters of CTCs positive for AR-V7 to guide treatment selection [58]. In a clever ap-
proach to better understand lung diseases, a microfluidic device was engineered to analyze indi-
vidual cell deformability as a means of diagnosing noninvasively whether human pleural effusions
were malignant or benign [70].

Although these devices can more readily isolate and analyze single cells, we are still left piecing to-
gether their implications in disease. Cytometry by time of flight (CyTOF), which can analyze the
contribution of a single cell to a given signaling pathway, addresses this point [71]. By using technol-
ogies such as CyTOF to assess complex populations of cells, the role of heterogeneity in driver mu-
tations and drug resistance can be comprehensively interrogated. In addition, longitudinal studies
need to be designed to decipher clone behavior, as well as spatial and temporal genomic signatures,
and to resolve evolutionary principles that underpin cancer as well as infectious diseases and other
disorders [72]. As a step towards determining the role of heterogeneity in disease progression and
treatment, a recent study performed single-cell triple-omic analysis of the genome, DNA methylome,
and transcriptome in a population of hepatocellular carcinoma cells to assess the larger-scale con-
tributions of genomic and epigenomic heterogeneity towards transcriptomic heterogeneity. These
capabilities in omics interrogation may provide extraordinary insight into disease heterogeneity and
enable both the prediction of drug responses and individualized therapy [73].

Sample collection poses many limitations for both personalized and precision medicine. Technologies
that can analyze small amounts of rare tissues or molecules will allow us to access information

Trends in Biotechnology, Month 2019, Vol. xx, No. xx 5

Trends in Biotechnology

Table 1. Clinicaltrials.gov Identifiers of Clinical Trials That Are Assessing Emerging Technologies for Personalized and Precision Medicine
Technology platform Status Key details of study Clinicaltrials.gov
identifier
Artificial intelligence Phase II Harnessing CURATE.AI to optimize N-of-1 combination therapy in NCT03759093
multiple myeloma.
Artificial intelligence Interventional Developing N-of-1 training trajectories via CURATE.AI. NCT03832101
Artificial intelligence Phase IV HIV: optimization of dose reduction of tenofovir (TDF) in antiretroviral NCT02632474
therapy (ART).
Artificial intelligence Phase II Liver and kidney transplant: dynamic modulation of therapeutic NCT03527238
dosing to optimize immunosuppression.
Artificial intelligence Interventional Radiotherapy: image-guided adaptive radiotherapy is being studied NCT04022018
to modulate intensity to reduce side effects and improve outcomes.
Genomic analysis/companion diagnostics Feasibility Non-small cell lung cancer (NSCLC): Co-clinical trials with genetic NCT02597738
mouse models, NSCLC patient specimens (plasma, serum, tissue),
and healthy patient samples to develop liquid biopsies.
Genome-guided therapy Observational Pain: assessment of the cytochrome P450 2D6 (CYP2D6) and NCT02664350
μ-opioid receptor gene (OPRM1) genotypes in guiding pain
management.
Genome-guided therapy Observational Preterm birth: genomic profiling to predict patient response to 17 NCT02173210
hydroxyprogesterone caproate (17OHPC) therapy to prevent
preterm birth.
Genome-guided therapy Phase II Breast cancer: biomarker analysis and genetic screening to develop NCT02624973
individualized treatment regimens.
Genome-guided therapy Phase II Multiple tumor types: matching of drugs to patients based on NCT02795156
genomic alterations.
Genome-guided therapy Phase III Prostate cancer: circulating tumor DNA is being used as a NCT03903835
biomarker for genome-guided drug selection.
Microfluidics/companion diagnostics Observational Multiple cancer types/circulating tumor cells: correlation of titer with NCT01022723
treatment response and progression for nasopharyngeal
carcinoma, breast cancer, colorectal cancer, prostate cancer, and
gastric cancer.
Microfluidics/companion diagnostics Observational Clear cell renal cancer/circulating tumor cells: comparison of NCT02499458
microfluidic technology that can detect CTCs under low epithelial
cell adhesion molecule (EpCAM) presence compared with
EpCAM-based detection platform for clear cell renal cancer.
Microfluidics/companion diagnostics Observational Metastatic breast cancer/circulating tumor cells: application of NCT02904135
antibody cocktails to capture CTCs from metastatic breast cancer
for microfluidic analysis.
Genome editing Phase I Mucopolysaccharidosis II (MPS II): intravenous delivery of SB-913, NCT03041324
a zinc finger nuclease, to enable production of iduronate 2-sulfatase
(IDS) enzyme from the liver.
Genome editing Phase I Mucopolysaccharidosis I (MPS I): intravenous delivery of SB-318, a NCT02702115
zinc finger nuclease, to insert α-L-iduronidase (IDUA) gene to
enable liver-mediated enzyme production.
Genome editing Phase I Severe hemophilia: intravenous delivery of SB-FIX, a zinc finger NCT02695160
nuclease, to insert the SB-FIX gene to enable Factor IX clotting
factor production.
Genome editing/cell therapy Phase I HIV: administration of zinc finger nuclease CCR5-modified NCT02500849
hematopoietic stem/progenitor cells (SB-728mR-HSPC) in patients
with undetectable disease/low CD4+ counts.
Nanomedicine/RNA therapy Phase 0 Glioblastoma: administration of NU-0129, a spherical nucleic acid NCT03020017
(SNA) therapy, to target the Bcl2L12 gene to arrest tumor growth.
Nanomedicine Phase I Cervical cancer: polysiloxane gadolinium chelates are delivered in NCT03308604
combination with cisplatin and radiation therapy to address locally
advanced disease.

6 Trends in Biotechnology, Month 2019, Vol. xx, No. xx

Trends in Biotechnology

Table 1. (continued)
Technology platform Status Key details of study Clinicaltrials.gov
identifier
Wearables Observational Perioperative risk assessment: pairing wearable (Garmin VivoSmart NCT03328039
HR+) with cardiopulmonary exercise testing indicators to develop
broadly applicable risk assessment protocol for high-risk elective
surgery patient population.
Wearables Observational Chronic obstructive airway disease (COPD): application of wearable NCT03268538
(on finger) for heart rate variability and oxygen saturation to detect
and manage COPD in patients over 60 years in age.
Wearables Observational Cardiology: a shirt-based electrocardiogram system is being used NCT03068169
to monitor the following patient cohorts: (i) cardiac rehabilitation
following myocardial infarct; (ii) pediatric superventricular
tachycardia; (iii) post-pulmonary vein isolation; (iv) cardiac
resynchronization therapy patients.
Digital Health Observational Alzheimer’s disease: cognitive battery software was used to predict NCT03676881
dementia onset.
Digital health Interventional Health eBrain study: application of neurocognitive assessment NCT02903862
mobile health software to assess brain health combined with
Mindoula intervention to address dementia and depression in
Alzheimer’s disease caregivers.
Biomaterials Phase II Wound healing: a nanodiamond-based biomaterial platform to NCT03376984
mediate periapical wound healing and prevention of reinfection
following root canal therapy.
Biomaterials Phase I 3D printing: personalized 3D printing of polycaprolactone (PCL) is NCT03348293
being explored for breast reconstruction after tumor removal.
Electronic health records/infrastructure/genotyping Observational Pain: CYP2D6 genotyping for pain management/control and NCT02335307
integration of pain questionnaire in electronic health record for
physician guidance.

previously lost in the bulk. Liquid biopsies, for instance, can be used to monitor disease in near
real-time and perhaps in the future be used to predict drug action and continuously modulate
combination therapy. Liquid biopsies could extract CTCs, exosomes, cell-free nucleic acids,
proteins, and other biological factors for collection and analysis [5,68,74]. This approach may
enable the identification of effective treatment outcomes and design criteria for drug-regimen
selection and predictors of response to the selected therapies.

Precision Therapeutics
Gene editing tools like clustered regularly interspaced short palindromic repeats
(CRISPR) and zinc finger nucleases (ZFN) have the ability to alter specific components of a
genome, thereby opening new doors to precision repair of defective genes in indications ranging
from cancer to HIV. The democratization of gene editing afforded by CRISPR promises to pro-
duce large amounts of data that we expect to shed light on new connections between diseases
and on new pathways never before implicated in a particular disease. Gene editing has recently
found its way into patients [75]. Recently, the ZFN-based genome editing therapeutic SB-913
was clinically administered to patients diagnosed with mucopolysaccharidosis II (MPS II), or
Hunter syndrome. One of the objectives of this study was to evaluate the impact of SB-913 on
leukocyte and plasma iduronate 2-sulfatase (IDS) enzyme activity in an effort to mediate lifelong
IDS production (NCT03041324). ZFN genome editing therapies are also being tested for MPS I
(SB-318, NCT02702115), and severe hemophilia B (SB-FIX, NCT02695160). Clinical trials eval-
uating CCR5-modified hematopoietic stem/progenitor cells (via ZFNs) in patients infected with
HIV-1 have also been initiated (NCT02500849).

Trends in Biotechnology, Month 2019, Vol. xx, No. xx 7

Trends in Biotechnology

Trends in Biotechnology

Figure 1. Engineering Precision Medicine Technology Platforms. From genome-guided medicine to clustered
regularly interspaced short palindromic repeats (CRISPR), a broad spectrum of technology platforms that bridge
engineering with precision medicine are poised to impact clinical outcomes.

Molecularly targeted and antibody therapies have served as cornerstones in precision medicine
and have resulted in a broad spectrum of approved drugs and emerging methods of patient-
specific drug prioritization via mutation databases [76–78]. Recent genome- and proteome-
guided therapeutic strategies that are in preclinical development have included the use of
CRISPR with an enzyme, CRISPR-associated protein-9 nuclease (Cas9), to correct genetic ab-
errations in diseases such as Duchenne muscular dystrophy (DMD). Specifically, CRISPR therapy
restored dystrophin production and substantially increased neuronal nitric oxide synthase (nNOS)
recruitment in a mouse model compared with the sham cohort, an important advance for
targeted gene editing [79]. Tumor profiling to assess the expression of programmed cell death
protein 1 (PD-1; which is encoded by the PDCD1 gene), PD-L1, and cytotoxic T lymphocyte-as-
sociated protein 4 (CTLA-4) from patient samples was used to predict treatment response for
melanoma [80]. Several classes of targeted inhibitors have been paired with genomic profiling
due to drug mechanisms of action that can specifically enhance treatment outcomes such as
progression-free survival. For example, epidermal growth factor receptor (EGFR) exon 19 dele-
tions and exon 21 point mutations served as indicators for subsequent afatinib administration
for non-small cell lung cancer (NSCLC) treatment in both treatment-naïve and refractory
NSCLC patients [81–83].

Administering CRISPR with improved efficacy can also be mediated using new delivery mecha-
nisms like nanoparticles. A recent preclinical study using CRISPR/Cas9 and a lipid nanoparticle
successfully switched off genes responsible for high cholesterol levels, in cells and mice [84].
Gold nanoparticle-mediated delivery of CRISPR-Cas9 was used to enhance the efficacy of
correcting a DNA mutation associated with DMD [85]. In addition to mediating guide RNA-
based therapy mediated by CRISPR, novel nanocarrier approaches are also being used for

8 Trends in Biotechnology, Month 2019, Vol. xx, No. xx

Trends in Biotechnology

RNA interference. For example, spherical nucleic acids (SNAs) are gold nanoparticles that
have been recently synthesized to be densely loaded with RNA duplexes for gene silencing. In
multiple in vitro and preclinical studies, SNAs were efficiently taken up in a broad spectrum of
cell lines and capable of mediating efficient RNA interference while remaining well tolerated [86–
88]. In a recent murine study, SNAs were utilized to knockdown the p53 inhibitor, Bcl2Like12
(Bcl2L12), which is overexpressed in glioblastoma, as a novel form of cancer treatment [89].
The SNA platform (NU-0129) was recently translated into the clinic, where a Phase 0 study is cur-
rently underway (NCT03020017).

Bioengineering for Personalized Medicine

Medicine is moving away from a ‘one size fits all’ mentality. We are realizing that patients with the
same disease can respond differently to everything, from drugs to biomaterials, and it is now time
to better understand this response and treat the patient as an individual. Engineers have the op-
portunity to make personalized medicine a greater reality in the clinic (Figure 2). Take biomaterials,
for instance. Researchers have found that a dextran-dendrimer composite works as an adhesive
differently in not only different organs, but also differently in the same organ under different envi-
ronments (e.g., colon cancer versus colitis) [90,91]. These findings suggest that the application of
most biomaterials cannot be generalizable and that the disease indication and organ environment
matters in the design of materials that are to reside temporarily within the body. Indeed, biomate-
rials can have a differential effect on cell fate, survival, and growth, yet there is a dearth of high-
throughput approaches to screen for the optimal material composition for a given application.
One approach is to use a combination of small and large animal models to determine the best
polymer carrier for islet transplantation [92–94]. An approach that avoids animal models alto-
gether looked at how different extracellular matrix formulations (different organs, different pro-
cessing methods) affected stem cell differentiation, cancer cell proliferation, and cell apoptosis
[95]. Engineering new platforms for narrowing down the optimal biomaterial formulation could im-
prove personalized biocompatibility and therapeutic outcomes.

Organ-on-a-chip platforms are being explored for potential individualized drug screening and tox-
icology studies. For example, a microfluidics-based model of human intestine was recently devel-
oped, where the complex gut microenvironment was recapitulated, allowing for the monitoring of
interactions between the gut microbiome, bacteria, and immune cells. The ability to replicate
organ-scale complexity using platforms such as this intestine model opens the doors to pharma-
cokinetics, absorption, and drug metabolism analysis as they relate to drug development and
toxicity studies [96,97]. These capabilities may also serve as the foundation for designing person-
alized treatments. To accurately personalize treatment, which includes parameters such as drug
selection and drug dosing, accurate and timely detection of treatment response is needed.
Conventional approaches to acquire these readouts include serum and urine analysis, or imaging
modalities such as ultrasound, CT, MRI, and x-rays, among others, which may be limited in terms
of testing frequency, especially if treatment response varies on a timescale of hours, or even mi-
nutes. Wearables and other classes of emerging technologies may overcome the challenges of
infrequent measurements to improve the accuracy of treatment response assessment, thereby
improving the design of personalized interventions [98]. As the field of personalized medicine con-
tinues to progress, new areas of development have included the use of nondrug-based plat-
forms, such as digital therapy, to address conditions such as mild cognitive impairment,
substance abuse, mental health, and attention deficit hyperactivity disorder, among others
[99–101].

A unifying attribute of these approaches is their potential towards using only a subject’s own data
to manage only their own care. This has been apparent with regards to AI-driven drug dosing as

Trends in Biotechnology, Month 2019, Vol. xx, No. xx 9

Trends in Biotechnology

Trends in Biotechnology

Figure 2. Engineering Personalized Medicine Technology Platforms. By bridging wearable technologies with artificial
intelligence and other engineering platforms, marked enhancements in the development of individualized treatment and
monitoring may be realized.

well as engineered cell therapy. Furthermore, a common attribute at the intersection of engineer-
ing and personalized medicine is that intervention and diagnosis can be adjusted in a serial
fashion for ongoing treatment optimization.

AI, Machine Learning, and Personalized Treatment

A major barrier to the optimization of any targeted therapy has been an inability to determine drug
doses that are best suited for an individual patient, especially since dosing requirements can often
change during the course of treatment. In the case of combination therapy, this is even more chal-
lenging given the virtually infinite dosing parameter space that exists. The same quandary also
confronts the concept of pinpointing population-optimized drug–dose ratios. To overcome this
challenge, conventional patient dosing and drug development has been guided by first identifying
the drugs to be administered, following by dose escalation-based trials to identify the maximum
tolerated dose (MTD). To increase the chances of successful treatment outcomes, drug-
synergy predictive modelling and other in silico methods have been employed [102]. However,
drug synergy and drug antagonism are dose-dependent, and not all genes involved in the uptake,
metabolism, and activity of any specific drug are known. Therefore, incorrect dosing can substan-
tially decrease or eliminate the efficacy of drug combinations, whereas optimized drug–dose ra-
tios can make normally ineffective drugs become potent when co-delivered.

In the drug and dose optimization arena, there remains a perception that correct dosing is impor-
tant solely for improving efficacy and safety. What is commonly misunderstood is that correct
dosing directly impacts the drugs that are selected for treatment in the first place [103,104].

10 Trends in Biotechnology, Month 2019, Vol. xx, No. xx

Trends in Biotechnology

Furthermore, in lieu of predictive drug discovery, synergy modeling, monotherapy, or combination

therapy selection, AI and machine learning (ML)-related platforms are now capable of determin-
istically identifying, in a continuous fashion, the drug administration parameters [9,105,106]. More
specifically, the implementation of novel AI-based platforms is precisely demonstrating the key re-
lationship between drug dose and drug selection by reconciling the virtually infinite parameter
space created by these factors. AI is also being leveraged to enhance imaging capabilities in the
area of diagnostics to further guide patient-specific treatment. In addition, the application of AI is
emerging in the area of therapeutics to simultaneously pinpoint the best drugs and doses, even
from very large pools of candidate compounds, for optimal combination therapy efficacy and safety
[107–118]. In a recent study using the deterministic AI-based quadratic phenotypic optimiza-
tion platform (QPOP) platform, novel drug combinations were identified from a pool of oncology
drug candidates without the need for synergy-based predictions or complex disease mechanism
information. In addition, QPOP was used to re-optimize the drug combination dose parameters
from the in vitro through preclinical stages of development. These AI-discovered combinations
substantially outperformed the clinical standards of care [117]. While QPOP implementation did
not require disease mechanism modelling, global optimization via this platform can lead to new in-
sights in the pathways that mediate globally optimized outcomes for comparison with conventional
target-based drug combination design. This could in turn lead to the addition of new drug candi-
dates in the combination therapy design pool. Approaches such as QPOP require marked reduc-
tions in data requirements as they are based on drug and dose inputs and quantifiable clinical
indicators of efficacy and safety (e.g., tumor burden, toxicity panels). This emerging capability of
mechanism-agnostic optimization, as well as a continuously increasing body of knowledge in the
genome-driven drug selection arenas are expected to address some of the barriers that have
been previously encountered with trials such as SHIVA, where molecular alterations serve as a
foundation for drug matching. In this previous study, the factors determining drug–patient pairing
were driven by an algorithm prioritizing molecular alterations and drug selection, among other fac-
tors. Lessons learned from this study, and suggestions provided by the clinical community, have
shed light on the potential of patient stratification using oligo-mutant tumors in lieu of multi-
mutant tumors, while additional suggestions included the use of combination therapy when
multiple molecular alterations were identified.

In combination therapy, the ‘sweet spot’ of an optimized drug–dose ratio and the best drugs that
comprise the multidrug treatment is almost always overlooked when dose escalation and MTD
are employed [10,106,119–124]. As such, the recent clinical application of augmented AI optimi-
zation platforms has overcome this challenge (NCT03759093, NCT03832101, NCT02632474,
NCT03527238) [105,106,119]. In a recent prospective trial, a new technology platform based
on PPM successfully personalized post-transplant immunosuppression [9]. The foundation of
this capability is based on the PPM-enabled discovery that drug administration (input) can be cor-
related to phenotypic output (e.g., tumor burden, bacterial/viral load) through a phenotypic re-
sponse surface. Without using algorithms or modelling, PPM optimization effectively calibrated
optimized dosing regimens for each patient enrolled in the study. Furthermore, patients were con-
tinuously calibrated to dynamically adjust their dosing guidelines for the entire duration of care, an
unprecedented capability [9,105]. Intrapatient dose-dependent drug synergy and antagonism
was also observed, further demonstrating the importance of optimizing personalized care.

These studies have since been expanded to include the optimization analysis of combination therapy
regimens in pediatric acute lymphoblastic leukemia, development of improved combination thera-
pies that maintain undetectable viral loads with substantial dose reduction in patients with HIV, as
well as optimization of immunosuppression dosing (NCT02632474, NCT03527238) [106].
Recently, CURATE.AI was harnessed to optimize N-of-1 combination therapy for an advanced

Trends in Biotechnology, Month 2019, Vol. xx, No. xx 11

Trends in Biotechnology

prostate cancer patient that resulted in a marked reduction in drug dosage that increased treatment
efficacy [125]. In addition, CURATE.AI was recently applied to the emerging field of digital therapeu-
tics, where N-of-1 profiles were constructed for cognitive training (NCT03832101) [99]. In addition to
AI-driven drug treatment, adaptive radiotherapy has also been explored, where images are being
used to modulate intensity to reduce treatment toxicity and enhance local disease control
(NCT04022018). These studies are examples of how AI can expand and optimize an arsenal of in-
terventions against many types of diseases.

As novel platforms for patient-specific testing and treatment continue to evolve, the implementa-
tion of AI is expected to expand. For example, organ-on-chip, single-cell interrogation platforms,
and powerful diagnostic modalities may represent ideal testbeds for N-of-1 or population-scale
drug development [126–133]. Specifically, organ-on-a-chip platforms are capable of replicating
complex function ranging from the gut, to placental transport and pulmonary function in both
healthy and diseased systems [96,97,127,134–136]. Recent advances in organ-on-a-chip devel-
opment have opened the doors to their applications in drug screening. Since combination ther-
apy is a mainstay for indications ranging from oncology to infectious diseases, among others,
these platforms can be integrated with AI-based screening strategies that reduce the number
of assays needed to identify optimized drug combinations [110,120]. In turn, this can accelerate
the clinical validation of novel treatment approaches.

Wearable and Implantable Sensors

Individualizing drug and dose parameters for indications ranging from pulmonary hypertension to
cancer therapy require constant and robust readout monitoring. To this end, wearable technology
will play a key role in mediating personalized therapy. Capacitive coupling-based flexible electronics
that do not require contact between cardiac tissue and metals were recently used for electrophys-
iological measurements in rabbits [137]. Wearable, battery-free devices for blood oxygenation and
heart rate measurements have also been developed [138,139]. Soft electronics for the analysis of
sweat, tears, skin interstitial fluid, and saliva link noninvasive measurements with blood concentra-
tions of these analytes [140–144]. These platforms can be used for glucose sensing and CF mon-
itoring without the barriers associated with fingerstick testing [145,146]. The clearance of the
MiniMed 670G (Medtronic), a closed-loop artificial pancreas, has enabled the algorithm-driven
delivery of basal insulin to control blood glucose levels for type 1 diabetes. Recent reporting of
real-world data revealed an increased time within the target glucose range with reduced incidence
of hyperglycemia and hypoglycemia [147]. Microfluidics-embedded polymer wristbands have also
been developed for pulse measurements [148]. Silicone microfiber tubes embedded with elec-
trodes for blood pressure, blood vessel stiffness, and heart rate monitoring have been developed,
allowing for noninvasive diagnosis and monitoring of atherosclerosis, venous ulcers, and potential
integration into bandages for diverse applications [149]. Such noninvasive, multiplexed monitoring
may ultimately identify personalized biomarker signatures [150].

Wearable devices are being tested in the clinic to potentially predict perioperative risk in patients un-
dergoing high-risk surgery by comparing wearable data (e.g., heart rate and exercise intensity) with
cardiopulmonary exercise testing standards (NCT03328039). Wearables are also being studied to
detect and manage chronic obstructive airway disease (NCT03268538). A wearable shirt-based
electrocardiogram measurement system can detect arrhythmias and could serve as a complemen-
tary real-time readout monitor for dynamically administered cardiovascular therapies (NCT03068169).

In the area of neurocognitive assessment, FDA-cleared mobile application software is also being
used to interrogate brain health for the development of interventional solutions in areas like demen-
tia, depression, and Alzheimer’s disease (AD) (NCT02903862, NCT03676881) [151]. Studies

12 Trends in Biotechnology, Month 2019, Vol. xx, No. xx

Trends in Biotechnology

employing digital medicine in clinical aging research are increasing since several methodological
barriers limit conventional in-clinic test administration. First, in-clinic testing is relatively infrequent.
In most studies, assessments typically occur on an annual basis in one extended testing session.
Second, day to day fluctuations in participant stressors, sleep patterns, and time of day effects,
and the artificial nature of testing environments, can strongly influence cognition and these sources
of variability may be exacerbated in participants with AD pathology [152]. Finally, the cognitive tests
most sensitive to the earliest declines in AD also demonstrate high levels of day-to-day variability
and can have substantial retest effects [153,154]. Smartphone assessment approaches afford
novel opportunities to overcome these barriers by leveraging the increasing adoption of
smartphones by older adults to mitigate the temporal, geographic, space, and personnel con-
straints imposed by in-person testing. Tests are administered in a measurement burst design, in
which a ‘burst’ of brief (30–60 seconds per test) cognitive tests are completed at random intervals
several times per day over the course of 1 week on participant’s personal smartphones. One study
‘visit’ thus represents the average performance across 7 days of assessments [155]. Advantages
include more frequent measurements at different times of day and a detailed evaluation of variability
and practice effects within a day and over a week, which may also serve as sensitive indicators of
disease stage [154,156,157]. As such, markedly increasing the frequency and scalability of data
collection using wearable and mobile health enhances data actionability, reducing the need to
use data that has been acquired using conventional approaches, and averaged over several
months. When paired with therapeutic strategies that have been properly tailored to this data,
wearables and smartphone-based approaches for continuous and robust health monitoring will
play a key role in mediating personalized therapy.

Data collection by wearables could allow for patients and their families to take action on their own
health and also avoid frequent trips to the doctor to have biomarker analyses performed using
conventional approaches. As these data troves grow, we can expect a major contribution beyond
personalized medicine to precision medicine, opening up new treatment ideas and options for
many patients.

Personalized Cell Therapy and Drug Delivery

The recent regulatory approval of chimeric antigen receptor T (CAR-T) cell immunotherapy
represents a major advance for personalized cancer treatment. The first of the approved CAR-T
therapies, tisagenlecleucel (Kymriah, Novartis), was developed to treat acute lymphoblastic leu-
kemia in patients 25 years old and younger, and involves removing a patient’s T cells, sending
them to a processing facility for reprogramming and expansion, and returning them to the doctor
for introduction to the patient [158]. Axicabtagene ciloleucel (Yescarta, KITE Pharma/Gilead
Sciences) was recently approved for the treatment of aggressive non-Hodgkin’s lymphomas,
including diffuse large B cell lymphoma [159,160]. A global effort to expand the indications that
are treated using CAR-T is underway. Furthermore, the regulatory approval of CAR-T therapy
represents a paradigm shift for the US FDA and a gateway towards the continued enhancement
of the efficacy and safety of living cell therapies. For example, recent studies have explored the
use of nonviral approaches, such as Sleeping Beauty transposition, which can enhance CAR-T
scalability for broader deployment [161]. This approach uses simple DNA minicircles to insert
CAR genes, potentially reducing the risk of mutagenesis and genotoxicity that may be associated
with viral modalities. This strategy may also reduce regulatory hurdles and the cost of CAR-T en-
gineering. Additional engineering approaches have sought to improve CAR-T manufacturing
strategies using ‘off-the-shelf’ cell therapy that does not require autologous T cells [162–164].
Through the use of CRISPR/Cas9 to eliminate CD7 and T cell receptor alpha chain, which are
also expressed on the malignant T cells, barriers such as fratricide could be avoided and pre-
served activity against T cell acute lymphoblastic leukemia in vitro and in vivo was demonstrated.

Trends in Biotechnology, Month 2019, Vol. xx, No. xx 13

Trends in Biotechnology

In addition, graft-versus-host-disease was not observed. In a recent advance, ZFN editing is

being harnessed to modify both autologous and allogeneic cell therapies, further broadening
the possibilities of off-the-shelf CAR-T manufacturing, potentially reducing the time to treatment
for the patient [165].

Cancer is not the only target of personalized cell therapy. Induced pluripotent stem cells taken from
a patient’s skin or elsewhere can be reprogrammed to a cell of choice, such as one of the kidney’s
specialized cells, the beta cell, to treat type 1 diabetes; or into a brain cell, which can be
transplanted into the body to, for instance, reduce inflammation in multiple sclerosis, as shown in
mice [93,166,167]. Additional examples of cell therapy include mitochondrial replacement
therapy (MRT), which has also been implemented in the UK. Mitochondrial disease is caused
when mutations in mitochondrial DNA (mtDNA) are maternally transferred to the offspring, which
can lead to serious disorders ranging from epilepsy to optic neuropathy and diabetes mellitus
and deafness, among others. To implement MRT, the healthy nucleus from the maternal egg
with malfunctioning mitochondria is transferred to a healthy egg (and donor mitochondria) without
a nucleus. In effect, this approach can result in a fertilized egg that contains nuclear DNA from two
parents and an mtDNA from a donor to eliminate genetic diseases in offspring [168].

The engineering of cells as biosensors is another particularly relevant area. Synthetic cells have
been designed for detecting cancer and for detecting diabetes indicators in urine [169,170].
Impressively, synthetic biology has advanced from bacteria to mammalian cells, with early bio-
sensors able to not only detect disease markers but also deliver therapeutic payloads to amelio-
rate symptoms. In this example, acute and chronic psoriasis was held in check by a population of
synthetic cells implanted in the mice [171]. This is the ultimate demonstration of personalized
medicine, where the cell is an autonomous sensor and therapy, delivering therapeutic levels of
‘drug’ without patient or clinician intervention.

Delivering therapy in a personalized fashion will also serve as an important enabling technology.
Knowing that dose modulation, or population optimized fixed dose combination therapy will serve
as a cornerstone for improving delivery technologies, methodologies such as personalized 3D printing
or biomaterial-mediated controlled release will play an increasingly important role in personalized
medicine (NCT03348293). Tailored release profiles from 3D printed tablets were recently introduced,
where the temporal drug release could be customized [172]. In addition, advances in micro-
manufacturing have produced drug-loaded 3D microstructures with temporal drug release control
using an approach termed stamped assembly of polymer layers (SEAL) [173]. Patient-specific
responses to combination therapy are often determined by unique dosing profiles. Therefore, these
tablets and microstructures may serve as viable drug delivery platforms for personalized medicine.

Combination therapy with nanomedicine has also received substantial attention [174–176].
Augmented AI was also used to design population-optimized combinatorial nanotherapy across
multiple breast cancer cell lines (MCF-7, MDA-MB-231, BT-20) and control cell lines (IMR-90,
MCF-10A, H9C2). This resulted in the identification of specific drug–dose ratios that mediated op-
timal treatment efficacy. Specifically, diverse dosing regimens of nanodiamonds (NDs)
functionalized with doxorubicin, bleomycin, and mitoxantrone (NDX, ND-BLEO, ND-MTX) and un-
modified paclitaxel were applied across the cell lines. Quadratic optimization, an AI-based ap-
proach that correlates drug inputs (e.g., drug and dose) and phenotypic outputs (e.g., cancer
cell death, control cell viability) using a smooth quadratic surface was conducted to accelerate
the identification of the dosages of the four therapies that mediated optimal cancer cell death
and control cell survival. Of note, The AI optimization showed that optimized ND-modified combi-
nation therapy outperformed ND-modified as well as unmodified single drug therapy, unmodified

14 Trends in Biotechnology, Month 2019, Vol. xx, No. xx

Trends in Biotechnology

combination therapy, as well as arbitrarily designed ND-modified combination therapy. While NDs
were used as the model delivery system, the augmented AI platform was universally applicable to
all classes of nanomaterials [120,177]. Clinical trials to validate ND-containing biomaterials have
since been initiated (NCT03376984) [178]. These studies show that the application of
engineering principles to address the barriers to achieving personalized medicine provides a strong
foundation for future developments in personalized materials, therapies, and devices.

Policy and Infrastructure Shifts in Personalized and Precision Medicine

The translation of approaches in personalized and precision medicine into the clinic and market-
place will depend on the evolution of the policy landscape and on the particular disease or indica-
tion, with some relying more on genomic/proteomic markers to guide treatment, and others
relying more on prognostic variables [179–181]. However, regardless of the indication being ad-
dressed, the transition of personalized and precision medicine into widespread use will likely chal-
lenge many long-held regulatory policies that are based on single products (rather than platforms)
and clinical standards, such as dose escalation, dose expansion, and MTD-based drug evalua-
tion and administration. In addition, the recent approval of vemurafenib was achieved using a bas-
ket study (MSK-IMPACT trial), where genomic drivers of the disease indication (Erdheim-Chester
disease) were used to assign treatment, as opposed to tumor location [182]. The opportunity to
regulate omics profile-specific medicines is likely imminent. Subpopulation-specific diagnostics
and therapies are now on our doorstep, suggesting that an emerging regulatory question will
be whether (and how) to integrate AI, ML, or other analytical platforms into clinical trial design.
In addition, as these platforms can be leveraged to individualize optimal treatment regimens, a
move beyond fixed P value-based evaluation may be needed due to the possibility of each patient
receiving an individualized regimen. These considerations open the door to potentially broaden-
ing the implementation of patient-centered clinical trials and the incorporation of Bayesian
decision analysis (BDA) and patient preference into clinical trial design. Trials based on BDA
may take into account the balance between benefits of the therapy or device itself on treatment
efficacy, risks of the treatment to the patient, and outcomes if the patient is not treated. These
and other factors may influence multiple aspects of the trial design. These include potentially re-
ducing the number of patients recruited to assess trial endpoints, biomarker-based patient selec-
tion, and statistical assessment approaches that are implemented, among others [183–185]. As
an example, in a recent study of CURATE.AI-based liver transplant immunosuppressant dosing,
small cohort statistical analysis was used to confirm that the AI-guided treatment cohort exhibited
reduced interpatient variance compared with control cohort patients. Ultimately, applying
approaches like AL, ML, and BCA may result in clinical trial designs that identify patient-specific
benefits that accelerate the approvals of novel therapies and devices.

The aforementioned opportunities to re-engineer the drug and device validation processes may
drive the evolution of regulatory science [186]. Next-generation regulatory architecture and
greater ‘regulatory literacy’ are needed. Streamlining the regulatory process for the rapidly evolv-
ing landscape of interventions and technologies will ultimately play a vital role in enabling person-
alized care to reach more patients in a timely fashion. Furthermore, issues such as drug pricing,
accessibility, repositioning in the context of regulating N-of-1 regimens, and other considerations
will need to be addressed [187]. The FDA has acknowledged that there could be a shift in focus
from primarily efficacy (because some studies have even been shown to be too efficacious!) to a
focus on long-term durability, safety, and product issues related to new technology, such as off-
target effects and implications longer-term.

The 21st Century Cures Act, which was signed into law in the United States at the end of 2016,
provides funding to the FDA to create new programs that will enhance its ability to expedite

Trends in Biotechnology, Month 2019, Vol. xx, No. xx 15

Trends in Biotechnology

approval of certain personalized and precision medicine products, such as cell therapies
(regenerative medicine advanced therapy) and medical devices (breakthrough devices). The
act goes beyond the FDA, also allocating money to the National Institutes of Health for medical
research and drug development. AI Singapore was recently unveiled to accelerate drug devel-
opment and enhance patient-centered care via more efficient hospital triaging processes
through the integration of AI and ML with advanced trial design and electronic medical
records (EMR). A partnership between the EU and European pharmaceutical companies
led to the recent creation of the Innovative Medicines Initiative (IMI). The IMI is combatting
major challenges that will confront the EU, including antimicrobial resistance, the need for con-
tinued flu vaccine development, and non-alcoholic fatty liver disease (NAFLD), a global issue
that affects as much as 30% of the EU population. Furthermore, NAFLD is capable of
transitioning to non-alcoholic steatohepatitis, which can lead to HCC. In addition, the IMI is
also developing novel tools to improve toxicity prediction in drug development as well as estab-
lishing a strategy to improve data quality during the preclinical drug development process in an
effort to shorten the pathway to approval and improve success rates [188].

Ten requirements for achieving a successful implementation of personalized and precision med-
icine were recently outlined [189]. Several of these have become increasingly relevant to the role
of engineering, in particular: (i) omics writ large, or the systematic and perpetual monitoring of en-
vironmental exposures that may drive disease risk; (ii) computation, which involves the integration
of patient EMR as a catalyst for clinical-decision support; and (iii) education, which plays a key role
in placing engineering at the foundation of future clinical leadership in personalized and precision
medicine. These requirements entail more frequent access to patients and innovative trial design.
Trial designs will increasingly need to focus on subpopulations of patients, and prespecifying
these patients for greater signals of efficacy will require a sea change in thinking about designs
that are tissue-agnostic and perhaps include a drug-diagnostic combination. It has been opined
that even the pathway for approving diagnostics may need to change (from the traditional 510k
route) in the face of modern diagnostics. In part, this may come through public awareness of
the efforts of precision medicine. For example, the Oncology Precision Network OpeN, started
in response to the Cancer Moonshot [rather than the Precision Medicine Initiative (PMI)], aims
to reach underserved patients and enroll them in clinical trials, as well as to foster data sharing to-
wards coordinated treatment decision support. Similarly, myriad websites, patient advocacy
groups, and nonprofits are geared up for greater patient engagement. Some believe that sharing
clinical trial results increases patient recruitment and engagement.

Study populations will need to be more diverse if we are to achieve the mission of precision med-
icine. Genome-wide association studies are still largely comprised of patients of European de-
scent [190]. In 2009, only 4% of patients in 373 studies were non-European; in 2016, only
19%, which is an improvement but still ‘persistent bias’ [191]. Therefore, re-engineering clinical
trial designs, standards, and protocols may have significant benefits for personalized and preci-
sion medicine. The Trans-Omics for Precision Medicine (TOPMed) Initiative launched by the US
National Heart, Lung, and Blood Institute aims to correct this course, as does its umbrella effort,
the NIH ‘All of Us’ Research Program (as part of the PMI); the Pharmacogenomic Resource for
Enhanced Decisions in Care and Treatment (PREDICT) program at Vanderbilt University
[192,193], the UK NHS counterpart, 100 000 Genomes Project; and other programs in China,
France, and elsewhere. Data sharing among these initiatives could also bolster the power of
these data sets to deliver precision medicine to all, rather than select populations that are overrep-
resented in individual precision medicine initiatives. In All of Us, engineers can play an important
role in helping to achieve its mission, for example, by developing mobile health devices for lifestyle
monitoring or designing technologies to discover new biomarkers of health and disease.

16 Trends in Biotechnology, Month 2019, Vol. xx, No. xx

Trends in Biotechnology

With regards to scaling the implementation of personalized and precision medicine in the clinic,
Outstanding Questions
the medical community is beginning to establish provider networks that are integrating genomic
What are the key technological barriers
information with electronic health records in order to enhance point-of-care guidance on diagnos- to the broader deployment of precision
tics and/or treatment. One example is the implementing genomics in practice (IGNITE) program and personalized medicine technologies?
[194–197], a network for the analysis of pharmacogenomic data and other family history informa-
What are the potential strategies that
tion for integration into EMR to power clinical decision support (CDS) and address a diverse can be employed to enhance the
spectrum of indications, such as hypertension renal disease and diabetes. Of note, a recent degree of integration of precision and
study pertaining to CYP2C19 genotype-guided antiplatelet therapy in a multisite trial from the personalized medicine technologies?
IGNITE network was reported [198]. Initiatives such as IGNITE will serve as foundations for the
Will effective clinical validation of
integration of personalized and precision medicine into mainstream healthcare (NCT02335307). precision and personalized medicine
technologies require new clinical trial
With respect to education policy, integrating engineering and clinical training will cultivate a gen- designs to be implemented?
eration of clinicians who are comfortable with emerging technologies. In turn, they will be well
When a patient’s therapeutic regimen
versed in clinical translation, a vital part of education in personalized and precision medicine. In is personalized only to their own
addition, the integration of healthcare economics, regulatory policy, and design of user interfaces physiology or disease characteristics
into programs dedicated to translational medicine will play a key role in driving the next generation during a clinical trial, how will statistical
analysis methods be adapted to
of technology transfer in personalized and precision medicine. Institutions that are addressing account for N-of-1, or single patient,
these issues include the Carle Illinois College of Medicine, which is placing engineering at the fore- studies?
front of clinical education; EnMed, a collaboration between Texas A&M University and Houston
Is the regulatory community prepared
Methodist Hospital; and The University of California, San Francisco (UCSF), which has developed
to utilize novel clinical trial designs,
a Masters in Translational Medicine program to bridge engineering and clinic-focused education where even a patient’s own regimens
to bring new technologies to the patient [199,200]. The integration of cellular and molecular engi- will constantly change over time, to bring
neering, modelling, simulation, and visualization as modalities for medical education, team-based new technologies towards widespread
implementation?
development of engineering solutions to unmet medical needs, as well as technology transfer are
examples of components that can be integrated into next generation curriculums to train the next How can innovators, educators,
generation of clinicians. These clinicians will ultimately have an enhanced grasp of the require- regulators, healthcare systems, and
ments for moving emerging technologies and platforms into humans, including regulation, intel- other stakeholders better prepare
themselves for the potential practice-
lectual property, and entrepreneurship, among other skillsets, and will ultimately accelerate changing impact of bringing precision
engineering-based discoveries into patient use [199]. and personalized medicine technolo-
gies into common clinical practice?

Concluding Remarks
Despite the obvious fact that patient physiology varies substantially from one individual to another,
drug development and patient care has largely relied on the administration of the same regimen to
an almost insurmountably diverse population. While certain scenarios permit bespoke drug admin-
istration according to constantly varying patient responses, arbitrary titration also remains the stan-
dard of care. These conventional treatment routes predispose patients towards suboptimal
response rates. The era of personalized and precision medicine will likely overcome this challenge.
Patients will no longer be confined to target-based drug selection and dose escalation-defined ad-
ministration protocols that collectively result in the standard one size fits all treatment approach.

The suite of enabling technologies for personalized and precision medicine has provided unprec-
edented abilities to identify and surveil disorders. These technologies can accurately diagnose
disease, comb large databases of omics information and EMRs to search for potential therapies,
identify biomarkers that may accurately reflect disease states, constantly monitor disease
progression or treatment response through wearable technology, and ultimately harness AI
and other optimization technologies to dynamically and accurately determine the best drugs
(chemotherapy, immunotherapy, nanotherapy, etc.) and corresponding doses that may change
as patient physiology evolves during the course of therapy. Achieving this level of treatment
personalization would require a seamless integration of biomarker development and potentially
a re-engineering of drug trial design and analysis paradigms. One challenge that arises when

Trends in Biotechnology, Month 2019, Vol. xx, No. xx 17

Trends in Biotechnology

this arsenal of enabling of technologies comes to fruition is the seamless integration of their
implementation (see Outstanding Questions).

The successful transitioning of validated precision and personalized medicine platforms into clin-
ical practice will require important advances beyond technology-centric innovation. A number of
risks and challenges associated with healthcare economics, ethics, and data privacy, which are
encountered after the stage of technology validation, need to be addressed [201–205]. In other
examples, rising healthcare costs across the globe are simultaneously creating a barrier and op-
portunity for the clinical implementation of novel technologies. Innovation through the lens of
payers, policymakers, will be vital. The minimum threshold to demonstrate patient and societal
value in healthcare will require improved treatment outcomes, reduced incidence of acute and
long-term treatment complications, and marked reductions in the cost of healthcare delivery
and administration. To address these criteria, a specific area of promise in both fundamental
and applied innovation is the use of AI-based approaches in drug development. As previously
mentioned, reconciling data-driven drug combination design into actionable regimens coupled
with attaining rapid fundamental insight into the pathways that drive the improved outcomes en-
abled by these regimens are examples of how engineering platforms for precision and personal-
ized medicine can move the needle in terms of impacting the practice of healthcare. With drug
prices reaching record levels, engineering approaches for precision and personalized medicine
will ultimately play a vital role in reducing drug development costs, increasing drug accessibility,
as well as contributing to cost-effectiveness for patients and healthcare systems, which itself is
a topic that may require further evaluation as precision and personalized medicine platforms
move towards broad deployability.

Mobilizing personalized and precision medicine for all will require a convergence of the aforemen-
tioned suite of enabling technologies and regulatory/public policy with advances in education and
coordinated efforts to deploy and fund truly personalized and precision medicine on a global scale.
At the core of this effort, biomedical engineering is playing an important role in catalyzing break-
throughs that will ultimately improve the human condition in an individualized fashion. Once we
gain a clearer picture of how these new capabilities can be paired and regulated to improve patient
outcomes, we can expect even more advances in personalized and precision medicine in the
present population, in underrepresented populations, and in future generations to come.

Acknowledgments
The authors acknowledge the important contributions of Dr Megan L. Frisk towards this work. We are deeply grateful for her
guidance, without which this work would not have been possible. We also thank Dr Vivian Y. Chang for her valuable expertise
pertaining to clinical genomics and the critical review of the manuscript. The authors also acknowledge the innovative discov-
eries in the personalized and precision medicine community that we were unable to include in this paper due to space lim-
itations. The authors also gratefully acknowledge Peter Wang and Theodore Kee.

Disclaimer Statement
D. Ho, E.K. Chow, X. Ding, C. Ho, and A. Zarrinpar are co-inventors of pending and issued patents pertaining to artificial
intelligence-based drug development.

References
1. Hamburg, M.A. and Collins, F.S. (2010) The path to personal- 4. Khoury, M.J. et al. (2003) Population screening in the age of
ized medicine. New Engl. J. Med. 363, 301–304 genomic medicine. New Engl. J. Med. 348, 50–58
2. Li, L. et al. (2015) Identification of type 2 diabetes subgroups 5. Sarioglu, A.F. et al. (2015) A microfluidic device for label-free,
through topological analysis of patient similarity. Sci. Transl. physical capture of circulating tumor cell clusters. Nat.
Med. 7, 311ra174 Methods 12, 685–691
3. Itakura, H. et al. (2015) Magnetic resonance image features 6. Miller, M.A. et al. (2015) Predicting therapeutic nanomedicine
identify glioblastoma phenotypic subtypes with distinct molec- efficacy using a companion magnetic resonance imaging
ular pathway activities. Sci. Transl. Med. 7, 303ra138 nanoparticle. Sci. Transl. Med. 7, 314ra183

18 Trends in Biotechnology, Month 2019, Vol. xx, No. xx

Trends in Biotechnology

7. Jonas, O. et al. (2015) An implantable microdevice to perform cancer diagnosis and hypoxia-activated cancer therapy. Adv.
high-throughput in vivo drug sensitivity testing in tumors. Sci. Ther. 2, 1800140
Transl. Med. 7, 284ra257 33. Unbehauen, M.L. et al. (2019) Tailor-made core-multishell
8. Ferguson, B.S. et al. (2013) Real-time, aptamer-based track- nanocarriers for the delivery of cationic analgesics to inflamed
ing of circulating therapeutic agents in living animals. Sci. tissue. Adv. Ther. 2, 1900007
Transl. Med. 5, 213ra165 34. Wang, C. et al. (2019) Cowpea mosaic virus promotes anti-
9. Zarrinpar, A. et al. (2016) Individualizing liver transplant immu- tumor activity and immune memory in a mouse ovarian tumor
nosuppression using a phenotypic personalized medicine plat- model. Adv. Ther. 2, 1900003
form. Sci. Transl. Med. 8, 333ra349 35. Tieu, T. et al. (2019) Advances in porous silicon–based
10. Lim, J.J. et al. (2019) Maximizing efficiency of artificial intelli- nanomaterials for diagnostic and therapeutic applications.
gence-driven drug combination optimization through minimal Adv. Ther. 2, 1800095
resolution experimental design. Adv. Ther. Published online 36. Lee, J. et al. (2019) Self-assembled aptamer nanoconstruct: a
October 28, 2019. https://doi.org/10.1002/adtp.201900122 highly effective molecule-capturing platform having therapeutic
11. Wang, Y. et al. (2018) Low-cost, μm-thick, tape-free electronic applications. Adv. Ther. 2, 1800111
tattoo sensors with minimized motion and sweat artifacts. NPJ 37. Yang, P. et al. (2019) Synthetic, supramolecular, and self-
Flex. Electron. 2, 6 adjuvanting CD8+ T-cell epitope vaccine increases the thera-
12. Koh, A. et al. (2016) A soft, wearable microfluidic device for the peutic antitumor immunity. Adv. Ther. 2, 1900010
capture, storage, and colorimetric sensing of sweat. Sci. 38. Leary, R.J. et al. (2010) Development of personalized tumor
Transl. Med. 8, 366ra165 biomarkers using massively parallel sequencing. Sci. Transl.
13. Rawson, T.M. et al. (2019) Microneedle biosensors for real-time, Med. 2, 20ra14
minimally invasive drug monitoring of phenoxymethylpenicillin: a 39. Gootenberg, J.S. et al. (2017) Nucleic acid detection with
first-in-human evaluation in healthy volunteers. Lancet Digit. Health CRISPR-Cas13a/C2c2. Science 356, 438–442
1, e335–e343 40. Ng, S-B. et al. (2017) EBV-associated primary nodal T/NK-cell
14. Pattichis, C.S. and Panayides, A.S. (2019) Connected health. lymphoma shows distinct molecular signature and copy num-
Front. Digit. Health 1, 1 ber changes. Haematologica 103, 278–287
15. van der Meel, R. et al. (2019) Smart cancer nanomedicine. Nat. 41. Ng, I.K. et al. (2017) Identifying large indels in targeted next
Nanotechnol. 14, 1007–1017 generation sequencing assays for myeloid neoplasms: a cau-
16. Neff, R.T. et al. (2017) BRCA mutation in ovarian cancer: test- tionary tale of the ZRSR1 pseudogene. J. Clin. Pathol. 70,
ing, implications and treatment considerations. Ther. Adv. 1069–1073
Med. Oncol. 9, 519–531 42. Chen, X-Q. et al. (2017) Dysregulation of neurotrophin signaling
17. Cheng, H.H. et al. (2017) Prostate cancer screening in a new in the pathogenesis of Alzheimer disease and of Alzheimer dis-
era of genetics. Clin. Genitourin. Cancer 15, 625–628 ease in Down syndrome. Free Radic. Biol. Med. 114, 52–61
18. Bednar, E.M. et al. (2017) A universal genetic testing initiative 43. Yong, W.P. et al. (2018) Real-time tumor gene expression pro-
for patients with high-grade, non-mucinous epithelial ovarian filing to direct gastric cancer chemotherapy: proof-of-concept
cancer and the implications for cancer treatment. Gynecol. “3G” trial. Clin. Cancer Res. 24, 5272–5281
Oncol. 146, 399–404 44. Mohan, S. et al. (2019) Profiling of circulating free DNA using
19. Yan, B. et al. (2017) Single-cell genomic profiling of acute my- targeted and genome wide sequencing in patients with small
eloid leukemia for clinical use: a pilot study. Oncol. Lett. 13, cell lung cancer. J. Thorac. Oncol. Published online October
1625–1630 16, 2019. https://doi.org/10.1016/j.jtho.2019.10.007
20. Pannone, L. et al. (2017) Structural, functional, and clinical 45. Montemayor, C. et al. (2019) Banking with precision:
characterization of a novel PTPN11 mutation cluster underlying transfusion medicine as a potential universal application in clin-
Noonan syndrome. Hum. Mutat. 38, 451–459 ical genomics. Curr. Opin. Hematol. 26, 480–487
21. Das, K., et al. Genomic predictors of chemotherapy efficacy in 46. Schubert, S.M. et al. (2016) Protein counting in single cancer
advanced or recurrent gastric cancer in the GC0301/TOP002 cells. Anal. Chem. 88, 2952–2957
phase III clinical trial. Cancer Lett. 412, 208-215. 47. Nie, S. et al. (2015) Correlations of salivary biomarkers with
22. Mohanty, A. et al. (2016) CCND1 mutations increase protein clinical assessments in patients with cystic fibrosis. PLoS
stability and promote ibrutinib resistance in mantle cell lym- One 10, e0135237
phoma. Oncotarget 7, 73558–73572 48. Wu, D., et al. Long-term measurements of human inflammatory
23. Smyth, E. et al. (2017) Pharmacogenetic analysis of the UK cytokines reveal complex baseline variations between individ-
MRC MAGIC trial: association of polymorphisms with toxicity uals. Am. J. Pathol. 187, 2620–2626.
and survival in patients treated with perioperative ECF chemo- 49. Kalinich, M. et al. (2017) An RNA-based signature enables high
therapy. Clin. Cancer Res. 23, 7543–7549 specificity detection of circulating tumor cells in hepatocellular
24. Weinshilboum, R.M. and Wang, L. (2017) Pharmacogenomics: carcinoma. Proc. Natl. Acad. Sci. 114, 1123–1128
precision medicine and drug response. Mayo Clin. Proc. 92, 50. Gawel, D.R. et al. (2019) A validated single-cell-based strategy
1711–1722 to identify diagnostic and therapeutic targets in complex dis-
25. Chia, S. et al. (2017) Phenotype-driven precision oncology as a eases. Genome Med. 11, 47
guide for clinical decisions one patient at a time. Nat. Commun. 51. Xu, J. et al. (2019) A new approach to find biomarkers
8, 435 in chronic fatigue syndrome/myalgic encephalomyelitis
26. Hawgood, S. et al. (2015) Precision medicine: beyond the in- (CFS/ME) by single-cell Raman micro-spectroscopy. Analyst
flection point. Sci. Transl. Med. 7, 300ps317 144, 913–920
27. Zhao, Z. et al. (2018) Transdermal immunomodulation: princi- 52. Kwok, S.J. et al. (2019) Multiplexed laser particles for spatially
ples, advances and perspectives. Adv. Drug Deliv. Rev. 127, resolved single-cell analysis. Light Sci. Appl. 8, 1–5
3–19 53. O’Kane, P.T. and Mrksich, M. (2017) An assay based on
28. Vela Ramirez, J.E. et al. (2017) Current state and challenges in SAMDI mass spectrometry for profiling protein interaction
developing oral vaccines. Adv. Drug Deliv. Rev. 114, 116–131 domains. J. Am. Chem. Soc. 139, 10320–10327
29. Yeo, D.C. et al. (2018) Abnormal scar identification with 54. Szymczak, L. et al. (2017) Peptide arrays: development and
spherical-nucleic-acid technology. Nat. Biomed. Eng. 2, application. Anal. Chem. 90, 266–282
227–238 55. Irish, J.M. et al. (2006) Mapping normal and cancer cell
30. He, L. et al. (2019) Polymer nanoplatforms at work in prostate signalling networks: towards single-cell proteomics. Nat. Rev.
cancer therapy. Adv. Ther. 2, 1800122 Cancer 6, 146–155
31. Mehta, P. et al. (2019) Broad scale and structure fabrication of 56. Gawad, C. et al. (2016) Single-cell genome sequencing: cur-
healthcare materials for drug and emerging therapies via rent state of the science. Nat. Rev. Genet. 17, 175–188
electrohydrodynamic techniques. Adv. Ther. 2, 1800024 57. Warren, L. et al. (2006) Transcription factor profiling in individ-
32. Zhang, X. et al. (2019) Supramolecular nanogel-based univer- ual hematopoietic progenitors by digital RT-PCR. Proc. Natl.
sal drug carriers formed by “soft–hard” co-assembly: accurate Acad. Sci. U. S. A. 103, 17807–17812

Trends in Biotechnology, Month 2019, Vol. xx, No. xx 19

Trends in Biotechnology

58. Scher, H.I. et al. (2016) Association of AR-V7 on circulating 83. Hochmair, M. et al. (2016) Complete remissions in afatinib-
tumor cells as a treatment-specific biomarker with outcomes treated non-small-cell lung cancer patients with symptomatic
and survival in castration-resistant prostate cancer. JAMA brain metastases. Anti-Cancer Drugs 27, 914–915
Oncol. 2, 1441–1449 84. Yin, H. et al. (2017) Structure-guided chemical modification of
59. Yu, M. et al. (2012) RNA sequencing of pancreatic circulating guide RNA enables potent non-viral in vivo genome editing.
tumour cells implicates WNT signalling in metastasis. Nature Nat. Biotechnol. 35, 1179–1187
487, 510–513 85. Lee, K. et al. (2017) Nanoparticle delivery of Cas9 ribonucleo-
60. Warrick, J. et al. (2010) A microfluidic cell concentrator. Anal. protein and donor DNA in vivo induces homology-directed
Chem. 82, 8320–8326 DNA repair. Nat. Biomed. Eng. 1, 889–901
61. Cima, I. et al. (2016) Tumor-derived circulating endothelial cell 86. Zheng, D. et al. (2012) Topical delivery of siRNA-based spher-
clusters in colorectal cancer. Sci. Transl. Med. 8, 345ra389 ical nucleic acid nanoparticle conjugates for gene regulation.
62. Zanini, F. et al. (2018) Single-cell transcriptional dynamics of Proc. Natl. Acad. Sci. U. S. A. 109, 11975–11980
flavivirus infection. eLife 7, e32942 87. Randeria, P.S. et al. (2015) siRNA-based spherical nucleic
63. Wang, W. et al. (2018) High fidelity hypothermic preservation of acids reverse impaired wound healing in diabetic mice by
primary tissues in organ transplant preservative for single cell ganglioside GM3 synthase knockdown. Proc. Natl. Acad.
transcriptome analysis. BMC Genomics 19, 140 Sci. U. S. A. 112, 5573–5578
64. Dalerba, P. et al. (2011) Single-cell dissection of transcriptional 88. Radovic-Moreno, A.F. et al. (2015) Immunomodulatory
heterogeneity in human colon tumors. Nat. Biotechnol. 29, spherical nucleic acids. Proc. Natl. Acad. Sci. U. S. A. 112,
1120–1127 3892–3897
65. Jan, M. et al. (2012) Clonal evolution of preleukemic hemato- 89. Jensen, S.A. et al. (2013) Spherical nucleic acid nanoparticle
poietic stem cells precedes human acute myeloid leukemia. conjugates as an RNAi-based therapy for glioblastoma. Sci.
Sci. Transl. Med. 4, 149ra118 Transl. Med. 5, 209ra152
66. Ritsma, L. et al. (2012) Intravital microscopy through an ab- 90. Artzi, N. et al. (2009) Aldehyde-amine chemistry enables mod-
dominal imaging window reveals a pre-micrometastasis stage ulated biosealants with tissue-specific adhesion. Adv. Mater.
during liver metastasis. Sci. Transl. Med. 4, 158ra145 21, 3399–3403
67. Ghajar, C.M. et al. (2013) The perivascular niche regulates 91. Oliva, N. et al. (2015) Regulation of dendrimer/dextran material
breast tumour dormancy. Nat. Cell Biol. 15, 807–817 performance by altered tissue microenvironment in inflamma-
68. Boffa, D.J. et al. (2017) Cellular expression of PD-L1 in the pe- tion and neoplasia. Sci. Transl. Med. 7, 272ra211
ripheral blood of lung cancer patients is associated with worse 92. Vegas, A.J. et al. (2016) Combinatorial hydrogel library enables
survival. Cancer Epidemiol. Biomark. Prev. 26, 1139–1145 identification of materials that mitigate the foreign body re-
69. Zhang, L. et al. (2016) Detection and characterization of circu- sponse in primates. Nat. Biotechnol. 34, 345–352
lating tumour cells in multiple myeloma. J. Circ. Biomark. 5, 10 93. Vegas, A.J. et al. (2016) Long-term glycemic control using
70. Tse, H.T.K. et al. (2013) Quantitative diagnosis of malignant polymer-encapsulated human stem cell-derived beta cells in
pleural effusions by single-cell mechanophenotyping. Sci. immune-competent mice. Nat. Med. 22, 306–311
Transl. Med. 5, 212ra163 94. Lind, J.U. et al. (2016) Instrumented cardiac microphysiological
71. Bendall, S.C. and Nolan, G.P. (2012) From single cells to deep devices via multimaterial three-dimensional printing. Nat.
phenotypes in cancer. Nat. Biotechnol. 30, 639–647 Mater. 16, 303
72. de Bruin, E.C. et al. (2014) Spatial and temporal diversity in 95. Beachley, V.Z. et al. (2015) Tissue matrix arrays for high-
genomic instability processes defines lung cancer evolution. throughput screening and systems analysis of cell function.
Science 346, 251–256 Nat. Methods 12, 1197–1204
73. Hou, Y. et al. (2016) Single-cell triple omics sequencing reveals 96. Bein, A. et al. (2018) Microfluidic organ-on-a-chip models of
genetic, epigenetic, and transcriptomic heterogeneity in hepa- human intestine. Cell. Mol. Gastroenterol. Hepatol. 5, 659–668
tocellular carcinomas. Cell Res. 26, 304–319 97. Prantil-Baun, R. et al. (2018) Physiologically based pharmaco-
74. Kenry et al. (2017) Single-layer ternary chalcogenide nanosheet kinetic nd pharmacodynamic analysis enabled by micro-
as a fluorescence-based “capture-release” biomolecular fluidically linked organs-on-chips. Annu. Rev. Pharmacol.
nanosensor. Small 13, 1601925 Toxicol. 58, 37–64
75. Ando, D. and Meyer, K. (2017) Gene editing: regulatory and 98. Blicharz, T.M. et al. (2018) Microneedle-based device for the
translation to clinic. Hematol. Oncol. Clin. North Am. 31, one-step painless collection of capillary blood samples. Nat.
797–808 Biomed. Eng. 2, 151–157
76. Fernández-Navarro, P. et al. (2019) The use of PanDrugs to 99. Kee, T., et al. Harnessing CURATE. AI as a digital therapeutics
prioritize anticancer drug treatments in a case of T-ALL platform by identifying N-of-1 learning trajectory profiles. Adv.
based on individual genomic data. BMC Cancer 19, 1005 Ther., 2, 1900023.
77. Russo, M. et al. (2019) Adaptive mutability of colorectal 100. Davis, N.O. et al. (2018) Proof-of-concept study of an at-home,
cancers in response to targeted therapies. Science. Published engaging, digital intervention for pediatric ADHD. PLoS One 13,
online November 7, 2019. https://doi.org/10.1126/science. e0189749
aav4474 101. Cho, C-H. and Lee, H-J. (2019) Could digital therapeutics be a
78. Fernandez-Nogueira, P. et al. (2019) Tumor associated fibro- game changer in psychiatry? Psychiatry Investig. 16, 97
blasts promote HER2-targeted therapy resistance through 102. Zimmer, A. et al. (2017) Prediction of drug cocktail effects when
FGFR2 activation. Clin. Cancer Res. Published online the number of measurements is limited. PLoS Biol. 15,
November 7, 2019. https://doi.org/10.1158/1078-0432. e2002518
CCR-19-0353 103. Zhang, J. et al. (2017) Integrating evolutionary dynamics into
79. Nelson, C.E. et al. (2016) In vivo genome editing improves treatment of metastatic castrate-resistant prostate cancer.
muscle function in a mouse model of Duchenne muscular Nat. Commun. 8, 1816
dystrophy. Science 351, 403–407 104. Gatenby, R.A. et al. (2009) Adaptive therapy. Cancer Res. 69,
80. Daud, A.I. et al. (2016) Tumor immune profiling predicts 4894–4903
response to anti–PD-1 therapy in human melanoma. J. Clin. 105. Chakradhar, S. (2017) Predictable response: finding optimal
Invest. 126, 3447–3452 drugs and doses using artificial intelligence. Nat. Med. 23,
81. Zenke, Y. et al. (2016) Successful treatment with afatinib after 1244
grade 3 hepatotoxicity induced by both gefitinib and erlotinib 106. Lee, D-K. et al. (2017) Optimizing combination therapy for
in EGFR mutation-positive non-small cell lung cancer. Lung acute lymphoblastic leukemia using a phenotypic personalized
Cancer 99, 1–3 medicine digital health platform. SLAS Technol. 22, 276–288
82. Venugopalan, A. et al. (2016) EGFR-targeted therapy results in 107. Wong, P.K. et al. (2008) Closed-loop control of cellular
dramatic early lung tumor regression accompanied by imaging functions using combinatory drugs guided by a stochastic
response and immune infiltration in EGFR mutant transgenic search algorithm. Proc. Natl. Acad. Sci. U. S. A. 105,
mouse models. Oncotarget 7, 54137–54156 5105–5110

20 Trends in Biotechnology, Month 2019, Vol. xx, No. xx

Trends in Biotechnology

108. Honda, Y. et al. (2013) Guiding the osteogenic fate of mouse 133. Titano, J.J. et al. (2018) Automated deep-neural-network sur-
and human mesenchymal stem cells through feedback system veillance of cranial images for acute neurologic events. Nat.
control. Sci. Rep. 3, 3420 Med. 24, 1337–1341
109. Xianting, D. et al. (2014) Discovery of a low order drug-cell re- 134. Huh, D. et al. (2010) Reconstituting organ-level lung functions
sponse surface for applications in personalized medicine. on a chip. Science 328, 1662–1668
Phys. Biol. 11, 065003 135. Cassidy, B. et al. (2018) Placental drug transport-on-a-chip: a
110. Mohd Abdul Rashid, M.B. et al. (2015) Identification and optimi- microengineered in vitro model of transporter-mediated drug
zation of combinatorial glucose metabolism inhibitors in hepa- efflux in the human placental barrier. Adv. Healthc. Mater. 7,
tocellular carcinomas. J. Lab. Autom. 20, 423–437 1700786
111. Weiss, A. et al. (2015) Rapid optimization of drug combinations 136. Miri, A.K. et al. (2018) Microfluidics-enabled multimaterial maskless
for the optimal angiostatic treatment of cancer. Angiogenesis stereolithographic bioprinting. Adv. Mater. 30, 1800242
18, 233–244 137. Fang, H. et al. (2017) Capacitively coupled arrays of multiplexed
112. Liu, Q. et al. (2015) Preclinical optimization of a broad-spectrum flexible silicon transistors for long-term cardiac electrophysiol-
anti-bladder cancer tri-drug regimen via the feedback system ogy. Nat. Biomed. Eng. 1, 0038
control (FSC) platform. Sci. Rep. 5, 11464 138. Kim, J. et al. (2017) Miniaturized battery-free wireless systems
113. Silva, A. et al. (2016) Output-driven feedback system control for wearable pulse oximetry. Adv. Funct. Mater. 27, 1604373
platform optimizes combinatorial therapy of tuberculosis using 139. Rogers, J.A. (2017) Nanomesh on-skin electronics. Nat.
a macrophage cell culture model. Proc. Natl. Acad. Sci. U. S. A. Nanotechnol. 12, 839
113, E2172–E2179 140. Tian, X. et al. (2019) Wireless body sensor networks based on
114. Lee, B-Y. et al. (2017) Drug regimens identified and optimized metamaterial textiles. Nat. Electron. 2, 243–251
by output-driven platform markedly reduce tuberculosis treat- 141. Cao, Y. et al. (2019) Self-healing electronic skins for aquatic en-
ment time. Nat. Commun. 8, 14183 vironments. Nat. Electron. 2, 75
115. Jha, S. and Topol, E.J. (2018) Information and artificial intelli- 142. Zhao, Y. et al. (2019) Design and applications of stretchable
gence. J. Am. Coll. Radiol. 15, 509–511 and self-healable conductors for soft electronics. Nano
116. Al-Shyoukh, I. et al. (2011) Systematic quantitative characteri- Converg. 6, 25
zation of cellular responses induced by multiple signals. BMC 143. Zhao, Y. et al. (2019) Highly conductive 3D metal-rubber
Syst. Biol. 5, 88 composites for stretchable electronic applications. APL Mater.
117. Rashid, M.B.M.A. et al. (2018) Optimizing drug combinations 7, 031508
against multiple myeloma using a quadratic phenotypic optimi- 144. Djenizian, T. et al. (2019) Advances in Flexible and Soft
zation platform (QPOP). Sci. Transl. Med. 10, eaan0941 Electronics, AIP Publishing
118. Ho, D. et al. (2019) Artificial intelligence in nanomedicine. 145. Gao, W. et al. (2017) Wearable physiological systems and
Nanoscale Horiz. 4, 365–377 technologies for metabolic monitoring. J. Appl. Physiol. 124,
119. Ho, D. et al. (2016) Diamonds, digital health, and drug develop- 548–556
ment: optimizing combinatorial nanomedicine. ACS Nano 10, 146. Emaminejad, S. et al. (2017) Autonomous sweat extraction and
9087–9092 analysis applied to cystic fibrosis and glucose monitoring using
120. Wang, H. et al. (2015) Mechanism-independent optimization of a fully integrated wearable platform. Proc. Natl. Acad. Sci. U. S. A.
combinatorial nanodiamond and unmodified drug delivery 114, 4625–4630
using a phenotypically driven platform technology. ACS Nano 147. Agrawal, P. et al. (2018) Real-world data from the MiniMed™
9, 3332–3344 670G system commercial launch. Diabetes 67, 960-P
121. Ho, D. et al. (2015) Nanodiamonds: the intersection of nano- 148. Gao, Y. et al. (2017) Wearable microfluidic diaphragm pressure
technology, drug development, and personalized medicine. sensor for health and tactile touch monitoring. Adv. Mater. 29,
Sci. Adv. 1, e1500439 1701985
122. Shen, Y., et al. Harnessing artificial intelligence to optimize long- 149. Xi, W. et al. (2017) Soft tubular microfluidics for 2D and 3D ap-
term maintenance dosing for antiretroviral-naive adults with plications. Proc. Natl. Acad. Sci. U. S. A. 114, 10590–10595
HIV-1 infection. Adv. Ther. Published online November 11, 150. Gao, W. et al. (2016) Fully integrated wearable sensor arrays for
2019. https://doi.org/10.1002/adtp.201900114. multiplexed in situ perspiration analysis. Nature 529, 509
123. Ding, X., et al. Harnessing an artificial intelligence platform to 151. Wallace, A. et al. (2015) Psychometric properties of a mobile
dynamically individualize combination therapy for treating colo- neurocognitive assessment tool. Alzheimers Dement. 11, P703
rectal carcinoma in a rat model. Adv. Ther. Published online 152. Bangert, A.S. and Balota, D.A. (2012) Keep up the pace: de-
November 20, 2019. https://doi.org/10.1002/adtp.201900127. clines in simple repetitive timing differentiate healthy aging
124. Ho, D. and Zarrinpar, A. (2017) Making N-of-1 Medicine a Re- from the earliest stages of Alzheimer’s disease. J. Int.
ality, SAGE Publications Neuropsychol. Soc. 18, 1052–1063
125. Pantuck, A.J. et al. (2018) Modulating BET bromodomain inhib- 153. Calamia, M. et al. (2013) The robust reliability of neuropsycho-
itor ZEN-3694 and enzalutamide combination dosing in a met- logical measures: meta-analyses of test–retest correlations.
astatic prostate cancer patient using CURATE. AI, an artificial Clin. Neuropsychol. 27, 1077–1105
intelligence platform. Adv. Ther. 1, 1800104 154. Hassenstab, J. et al. (2015) Absence of practice effects in pre-
126. Guan, A. et al. (2017) Medical devices on chips. Nat. Biomed. clinical Alzheimer’s disease. Neuropsychology 29, 940–948
Eng. 1, 0045 155. Rast, P. et al. (2012) Intensive measurement designs for re-
127. Zhang, Y.S. et al. (2017) Multisensor-integrated organs-on- search on aging. GeroPsych 25, 45–55
chips platform for automated and continual in situ monitoring 156. Sliwinski, M.J. (2008) Measurement-burst designs for social
of organoid behaviors. Proc. Natl. Acad. Sci. U. S. A. 114, health research. Soc. Personal. Psychol. Compass 2, 245–261
E2293–E2302 157. Duchek, J.M. et al. (2013) Relationship between Stroop perfor-
128. Shalek, A.K. and Benson, M. (2017) Single-cell analyses to tai- mance and resting state functional connectivity in cognitively
lor treatments. Sci. Transl. Med. 9, eaan4730 normal older adults. Neuropsychology 27, 516–528
129. Ting, D.S. et al. (2018) AI for medical imaging goes deep. Nat. 158. Prasad, V. (2017) Tisagenlecleucel-the first approved CAR-T-
Med. 24, 539 cell therapy: implications for payers and policy makers. Nat.
130. Ardila, D. et al. (2019) End-to-end lung cancer screening with Rev. Clin. Oncol. 15, 11–12
three-dimensional deep learning on low-dose chest computed 159. Roberts, Z.J. et al. (2017) Axicabtagene ciloleucel, a first-in-
tomography. Nat. Med. 25, 954 class CAR T cell therapy for aggressive NHL. Leuk. Lymphoma
131. Ngiam, K.Y. and Khor, W. (2019) Big data and machine learn- 59, 1785–1796
ing algorithms for health-care delivery. Lancet Oncol. 20, 160. No authors listed. FDA Approves Second CAR T-cell Therapy.
e262–e273 Cancer Discov. 8, 5–6
132. Coudray, N. et al. (2018) Classification and mutation prediction 161. Monjezi, R. et al. (2016) Enhanced CAR T-cell engineering
from non-small cell lung cancer histopathology images using using non-viral Sleeping Beauty transposition from minicircle
deep learning. Nat. Med. 24, 1559 vectors. Leukemia 31, 186

Trends in Biotechnology, Month 2019, Vol. xx, No. xx 21

Trends in Biotechnology

162. Sadelain, M. et al. (2017) Therapeutic T cell engineering. Nature 184. Das, S. and Lo, A.W. (2017) Re-inventing drug development: a
545, 423 case study of the I-SPY 2 breast cancer clinical trials program.
163. Cooper, M.L. et al. (2018) An “off-the-shelf” fratricide-resistant Contemp. Clin. Trials 62, 168–174
CAR-T for the treatment of T cell hematologic malignancies. 185. Wong, C.H. et al. (2018) Estimation of clinical trial success rates
Leukemia 32, 1970–1983 and related parameters. Biostatistics 20, 273–286
164. Ruella, M. and Kenderian, S.S. (2017) Next-generation chimeric 186. Altman, R.B. et al. (2016) A research roadmap for next-
antigen receptor T-cell therapy: going off the shelf. BioDrugs generation sequencing informatics. Sci. Transl. Med. 8,
31, 473–481 335ps310
165. No authors listed. Kite, Sangamo partner on gene-edited cell 187. van der Gronde, T. et al. (2017) Addressing the challenge of
therapies. Cancer Discov. 8, 379–380 high-priced prescription drugs in the era of precision medicine:
166. Peruzzotti-Jametti, L. et al. (2018) Macrophage-derived extra- a systematic review of drug life cycles, therapeutic drug mar-
cellular succinate licenses neural stem cells to suppress kets and regulatory frameworks. PLoS One 12, e0182613
chronic neuroinflammation. Cell Stem Cell 22, 355–368 188. Goldman, M. (2012) The innovative medicines initiative: a
167. Ma, H. et al. (2018) Establishment of human pluripotent stem European response to the innovation challenge. Clin.
cell-derived pancreatic β-like cells in the mouse pancreas. Pharmacol. Ther. 91, 418–425
Proc. Natl. Acad. Sci. U. S. A. 115, 3924–3929 189. Kohane, I.S. (2015) Ten things we have to do to achieve preci-
168. Saxena, N. et al. (2018) Mitochondrial donation: a boon or sion medicine. Science 349, 37–38
curse for the treatment of incurable mitochondrial diseases. 190. Rosenberg, N.A. et al. (2010) Genome-wide association stud-
J. Hum. Reprod. Sci. 11, 3–9 ies in diverse populations. Nat. Rev. Genet. 11, 356–366
169. Danino, T. et al. (2015) Programmable probiotics for detection 191. Popejoy, A.B. and Fullerton, S.M. (2016) Genomics is failing on
of cancer in urine. Sci. Transl. Med. 7, 289ra284 diversity. Nature 538, 161–164
170. Courbet, A. et al. (2015) Detection of pathological biomarkers in 192. Carlson, B. (2012) Vanderbilt pioneers bedside genetics.
human clinical samples via amplifying genetic switches and Biotechnol. Healthc. 9, 31
logic gates. Sci. Transl. Med. 7, 289ra283 193. Wells, Q.S. et al. (2019) Accelerating biomarker discovery
171. Schukur, L. et al. (2015) Implantable synthetic cytokine con- through electronic health records, automated biobanking,
verter cells with AND-gate logic treat experimental psoriasis. and proteomics. J. Am. Coll. Cardiol. 73, 2195–2205
Sci. Transl. Med. 7, 318ra201 194. Weitzel, K.W. et al. (2016) The IGNITE network: a model for ge-
172. Sun, Y. and Soh, S. (2015) Printing tablets with fully customiz- nomic medicine implementation and research. BMC Med.
able release profiles for personalized medicine. Adv. Mater. 27, Genet. 9, 1–13
7847–7853 195. Orlando, L.A. et al. (2017) Developing a common framework for
173. McHugh, K.J. et al. (2017) Fabrication of fillable microparticles evaluating the implementation of genomic medicine interven-
and other complex 3D microstructures. Science 357, tions in clinical care: the IGNITE Network’s Common Measures
1138–1142 Working Group. Genet. Med. 20, 655–663
174. Tai, W. et al. (2014) Folding graft copolymer with pendant drug 196. Sperber, N.R. et al. (2017) Challenges and strategies for
segments for co-delivery of anticancer drugs. Biomaterials 35, implementing genomic services in diverse settings: experi-
7194–7203 ences from the Implementing GeNomics In pracTicE (IGNITE)
175. von Maltzahn, G. et al. (2011) Nanoparticles that communicate network. BMC Med. Genet. 10, 35
in vivo to amplify tumour targeting. Nat. Mater. 10, 545–552 197. Cavallari, L.H. et al. (2017) The IGNITE Pharmacogenetics
176. Deng, Z.J. et al. (2013) Layer-by-layer nanoparticles for sys- Working Group: an opportunity for building evidence with phar-
temic codelivery of an anticancer drug and siRNA for potential macogenetic implementation in a real-world setting. Clin.
triple-negative breast cancer treatment. ACS Nano 7, Transl. Sci. 10, 143–146
9571–9584 198. Cavallari, L.H. et al. (2017) Multisite investigation of outcomes
177. Ding, X. et al. (2017) Effective drug combination for with implementation of CYP2C19 genotype-guided antiplatelet
Caenorhabditis elegans nematodes discovered by output- therapy after percutaneous coronary intervention. J. Am. Coll.
driven feedback system control technique. Sci. Adv. 3, Cardiol. Intv. 11, 181–191
eaao1254 199. Chien, S. et al. (2015) Engineering as a new frontier for transla-
178. Lee, D-K. et al. (2017) Clinical validation of a nanodiamond- tional medicine. Sci. Transl. Med. 7, 281fs213
embedded thermoplastic biomaterial. Proc. Natl. Acad. Sci. 200. Kurpinski, K. et al. (2014) Mastering translational medicine:
U. S. A. 114, E9445–E9454 interdisciplinary education for a new generation. Sci. Transl.
179. Blaus, A. et al. (2015) Personalized cardiovascular medicine Med. 6, 218fs212
today. A Food and Drug Administration/Center for Drug Evalu- 201. Reddy, S. et al. (2019) A governance model for the application
ation and Research perspective. Circulation 132, 1425–1432 of AI in health care. J. Am. Med. Inform. Assoc. Published on-
180. Vicini, P. et al. (2016) Precision medicine in the age of big data: line November 4, 2019. https://doi.org/10.1093/jamia/ocz192
the present and future role of large-scale unbiased sequencing 202. Cohen, J. (2019) Did CRISPR help—or harm—the first-ever
in drug discovery and development. Clin. Pharmacol. Ther. 99, gene-edited babies? Science Published online August 1,
198–207 2019. https://doi.org/10.1126/science.aay9569
181. Fang, H. et al. (2016) FDA drug labeling: rich resources to facil- 203. Dinh-Le, C. et al. (2019) Wearable health technology and elec-
itate precision medicine, drug safety, and regulatory science. tronic health record integration: scoping review and future di-
Drug Discov. Today 21, 1566–1570 rections. JMIR mHealth uHealth 7, e12861
182. Hyman, D.M. et al. (2015) Vemurafenib in multiple 204. Tafuri, W.L. et al. (1988) Dogs as experimental models for the
nonmelanoma cancers with BRAF V600 mutations. N. Engl. study of the natural course of Chagas disease. Rev. Soc.
J. Med. 373, 726–736 Bras. Med. Trop. 21, 77
183. Chaudhuri, S.E. et al. (2018) Patient-centered clinical trials. 205. Lee, S.S-J. et al. (2019) Ethics of inclusion: cultivate trust in pre-
Drug Discov. Today 23, 395–401 cision medicine. Science 364, 941–942

22 Trends in Biotechnology, Month 2019, Vol. xx, No. xx